Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Assessing the reliability of the 1760 British geographical survey of the St. Lawrence River Valley
(USC Thesis Other)
Assessing the reliability of the 1760 British geographical survey of the St. Lawrence River Valley
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Assessing the Reliability of the 1760 British Geographical Survey of the St. Lawrence River Valley
by
Nicholas Gliserman
A Thesis Presented to the
Faculty of the USC Graduate School
University of Southern California
In Partial Fulfillment of the
Requirements for the Degree
Master of Science
(Geographic Information Science and Technology)
May 2018
ii
Copyright © 2018 by Nicholas Gliserman
iii
To my loving wife, Christina Copland
iv
Table of Contents
List of Figures ................................................................................................................................. vi
List of Tables ................................................................................................................................... x
Acknowledgements ........................................................................................................................ xi
List of Abbreviations .................................................................................................................... xii
Abstract ........................................................................................................................................ xiii
Chapter 1: Introduction .................................................................................................................... 1
Chapter 2: Towards a Historiographical Synthesis of Archival and Digital Methods .................... 7
2.1. The Underwhelming Presence of Historical Maps in the Spatial Humanities ................... 8
2.2. Scholarship on the Murray Atlas ...................................................................................... 12
Chapter 3: Historical Overview of Cartography in Canada during the French Regime
and through the British Conquest ........................................................................................... 18
3.1. The French Mapping of Canada and the Shaping of the Human Geography ................... 18
3.2. British Mapmaking and the Invasion of Canada during the Seven Years War ................ 28
3.3. Surveying, Drafting, and Gifting The Murray Atlas ........................................................ 31
Chapter 4: Digitally Processing the Murray Atlas ........................................................................ 51
4.1. Creating a Basemap and Vector Data ............................................................................... 51
4.2. Spatial Adjustment ........................................................................................................... 60
4.3. Spatial Distortion .............................................................................................................. 65
4.4. Boundaries of Enumeration .............................................................................................. 74
Chapter 5: The Reliability of the Murray Atlas as a Demographic Source ................................... 82
5.1. Historical Context of Early Canadian Demographic Sources .......................................... 82
5.2. Potential Pitfalls in Working with the Census and Parish Registers ................................ 90
5.3. Assessing the Quality of the Censuses ............................................................................. 98
v
5.4. The Parish Data and Murray Atlas as Demographic Sources that Reinforce
Each Other .................................................................................................................. 106
Chapter 6: Conclusions ................................................................................................................ 113
References ................................................................................................................................... 122
Appendix A : Models for Reconstructing Missing Parish Records ............................................ 127
Appendix B : Modeling Population in Canada with Parish Registers ......................................... 133
Appendix C : Comparing 1760 and 1765 Censuses by Region .................................................. 137
Appendix D : Changes in Population According to 1762 and 1765 Censuses ........................... 138
Appendix E : Hypothetical Colony-Wide CBRs for 1760 Census based on Average
Family Size ..................................................................................................................... 139
Appendix F : Regional Distribution of Dwellings in Murray Atlas (with Extrapolated
Population and CBR) by Region. .................................................................................... 140
vi
List of Figures
Figure 1: “PLAN of the Settled Part of the Province of Quebec Reduced from the large
Survey to serve as an Index Plan To the Large Sheets” (c.1762) ........................................... 2
Figure 2: Study Area. St. Lawrence River Valley with the Superimposed Sheets of the
Murray Atlas (c.1762). ............................................................................................................ 3
Figure 3: Detail of Beauport from Shields, “Cartometric Assessment of the Quebec City
Region” in “Murray Map Cartographically Considered” ..................................................... 15
Figure 4: Detail of Beauport in the Murray Atlas. ........................................................................ 16
Figure 5: Jean Bourdon “Carte depuis Kébec jusque au cap de Tourmente” (1641). .................. 21
Figure 6: Jean Bourdon, “Vray Plan du haut & bas de quebec comme il est en l[’]an 1660”
(c.1660). ................................................................................................................................ 22
Figure 7: Robert de Villeneuve, “Plan De La Ville Et Chateau de Quebec, Fait En 1685,
Mezvrée Exactement” (1685). .............................................................................................. 24
Figure 8: Villeneuve, “CARTE DES-ENUIRONS DE QVEBEC EN LA NOUVELLE
FRANCE Mezuré Sur le Lieu Tres-Exactement, En 1685, Et 86, PAR LE S
r
DEVILLENEVVE Injenieur du Roy” (1685-6) ................................................................... 25
Figure 9: Gédéon de Catalogne and Jean Baptiste de Couagne, “Carte du gouvernement
des trois rivieres” (1709) ....................................................................................................... 26
Figure 10: The walled city of Trois Rivieres, surrounded by long, narrow lots. .......................... 26
Figure 11: Samuel Holland and [John] Montresor, “A PLAN of the RIVER St.
LAURENCE from MONTREAL TO BERTHEIR ON THE NORTH and to SOREL
ON THE SOUTH SIDE OF THE RIVER being part OF A SURVEY INTENDED
TO QUEBEC” (c.1761) ........................................................................................................ 34
Figure 12: Dedication and Coat of Arms in Pitt’s copy of the Murray Atlas ............................... 34
Figure 13: Terrain North of the Island of Bouchard in Pitt’s copy. .............................................. 35
Figure 14: Same area as in Figure 13, depicted in Murray’s personal copy of the Murray
Atlas ...................................................................................................................................... 36
Figure 15: Cultivated land around St. Sulpice in Pitt’s copy ........................................................ 36
Figure 16: Same area as in Figure 15, depicted in Murray’s personal copy ................................. 37
Figure 17: Present-day aerial photography of the same area in Figure 13 and Figure 14 ............ 37
vii
Figure 18: [Murray et al] “[Map in two sheets of the inhabited parts of Canada]” (c.1762). ...... 40
Figure 19: [Murray et al], “[Map of Quebec and the environs, with plan of military
operations of 1759]” (c.1762) ............................................................................................... 40
Figure 20: Detail in Figure 19. ...................................................................................................... 41
Figure 21: Detail from Figure 18. ................................................................................................. 42
Figure 22: An example of notes about parishes in the King’s copy ............................................. 44
Figure 23: Inset in the King’s copy of the Murray Atlas .............................................................. 47
Figure 24: Detail from Pitt’s copy. ............................................................................................... 48
Figure 25: Assembling the Individual Maps on top of the Composite Map in ArcGIS ............... 52
Figure 26: Example of Insets in the Murray Atlas ........................................................................ 53
Figure 27: Example of Mapped Area Extending beyond Neatline ............................................... 53
Figure 28: Assembling Insets in Photoshop ................................................................................. 54
Figure 29: Inset Alignment with the Composite Map .................................................................. 55
Figure 30: Aligning the Neatline with the Grid in Case with Overhang ...................................... 55
Figure 31: Murray Atlas Entity Relationship Diagram (ERD) ..................................................... 57
Figure 32: Creating Vector Data in ArcGIS by Tracing Over the Historical Map ....................... 58
Figure 33: Distribution of Control Points for Spatial Adjustment ................................................ 61
Figure 34: The Digitized Murray Atlas ........................................................................................ 62
Figure 35: View of Quebec and Environs ..................................................................................... 63
Figure 36: View of Montreal and Environs .................................................................................. 64
Figure 37: Linear Distortion in the Murray Atlas. ........................................................................ 66
Figure 38: Linear Distortion Greater than 10% in the Murray Atlas ............................................ 66
Figure 39: Linear Error Greater than 10% in the Eastern Portion of the Murray Atlas ............... 67
Figure 40: Row from the Rightmost Portion of the Composite Atlas. ......................................... 67
Figure 41: Absence of Linear Distortion Above 10% from Beauport and Île d’Orleans ............. 69
Figure 42: The Relationship between Surveyor Coverage and Distortion ................................... 70
viii
Figure 43: Surveyor Coverage and Error in the Murray Atlas ..................................................... 71
Figure 44: Error around Lake St. Peter with Dwellings in Green ................................................ 72
Figure 45: Points to Sample Linear Error and Clustering ............................................................. 73
Figure 46: Scatterplot Showing Clustering vs. Linear Error ........................................................ 73
Figure 47: Scatterplot Showing Density vs. Linear Error ............................................................. 74
Figure 48: Some Boundary Options. ............................................................................................ 77
Figure 49: Creating Region Boundaries using Thiessen Polygons ............................................... 78
Figure 50: Creating Region Boundaries using Natural Barriers. .................................................. 78
Figure 51: Documenting Sources and Rationales for Boundaries in the Geodatabase. ................ 80
Figure 52: Governments and Regions of the St. Lawrence River Valley, c.1760-1765 ............... 81
Figure 53: Page from Roll of Names for the Census of Canada (1666). ...................................... 84
Figure 54: “Recensement général du Canada” (1685) .................................................................. 85
Figure 55: François Plouf’s baptism record from the Contrecoeur parish register ....................... 88
Figure 56: PRDH’s “Individual File" ........................................................................................... 90
Figure 57: “A General Return for the Government of Montreal” in Thomas Gage, “[State
of Government of Montreal]” (March 20, 1762). ................................................................. 93
Figure 58: “Number of Souls in the Government of Quebec” in “[Report on the
Government of Quebec]” (5 June 1762). .............................................................................. 94
Figure 59: Number of Parishes vs. Surviving Parish Records, 1615-1765 .................................. 96
Figure 60: Percentage of Surviving Parish Records, 1615-1765. ................................................. 97
Figure 61: Survival of Parish Registers by Region ....................................................................... 97
Figure 62: Demographic Change According to the 1760, 1762, and 1765 Censuses .................. 99
Figure 63: Raw Number Births in the Government of Quebec, 1750-1765. .............................. 101
Figure 64: CBR According to A) Population Projections from Dwellings in the Murray
Atlas and B) 1760 Parish Registers .................................................................................... 108
Figure 65: Migration as a Percent of Natural Growth ................................................................ 110
ix
Figure 66: Relationship between Time and Net Migration ........................................................ 111
Figure 67: Dwellings within and beyond 1 Kilometer of a Major Waterway ............................ 116
Figure 68: Kernel Density Analysis of Dwellings. ..................................................................... 118
Figure 69: Kernel Density Analysis of Dwellings in Montreal. ................................................. 119
Figure 70: Regional Distribution of Projected Natural Population Growth by Decade. ............ 132
Figure 71: Canadian Population, 1615-1765 .............................................................................. 134
Figure 72: Share of Canadian Population from Natural Growth and Immigration, 1615-
1765..................................................................................................................................... 134
Figure 73: CBR Calculated from Only Surviving Registers vs. Surviving Registers with
Estimated Values for Non-Surviving Registers, 1616-1651. .............................................. 135
Figure 74: CBR Calculated from Surviving Registers Only vs. Surviving Registers and
Estimated Values for Non-Surviving Registers, 1645-1765. .............................................. 136
x
List of Tables
Table 1: Range of Error in Quadrat Lines ..................................................................................... 65
Table 2: Distribution of Error by Government .............................................................................. 68
Table 3: Surveyor Coverage and Error .......................................................................................... 70
Table 4: Distribution of Recorded Changes in Number of Families between 1760 and 1765
Censuses. ................................................................................................................................ 99
Table 5: Raw number of births in Quebec, 1755-1765, according to surviving Parish
Registers. .............................................................................................................................. 101
Table 6: Distribution of Recorded Changes in Population between 1762 and 1765 Censuses ... 102
Table 7: Summary of Hypothetical Colony-Wide CBRs for 1760 Census based on Average
Family Size .......................................................................................................................... 105
Table 8: Description of Models to Estimate Missing Parish Values ........................................... 128
Table 9: Model Error for Estimating the Values of Missing Parish Registers ............................ 131
Table 10: Population Estimates and Source Comparisons .......................................................... 133
xi
Acknowledgements
Usually a Master’s degree precedes the Doctorate—but in this case, the trajectory was the
opposite. A dissertation on the history of cartography in North America sparked my interest in
GIS as a handmaiden to historical scholarship. So, in the sixth year of my Ph.D., I began taking
courses at USC’s Spatial Sciences Institute where I was quite lucky to find strong mentorship
coupled with a new and exciting way of thinking about my dissertation topic.
I am grateful to my Ph.D. advisor Peter Mancall for letting me pursue this whim and to
my teachers at the Spatial Sciences Institute. Here Jordan Hastings and my advisor Karen Kemp
deserve special thanks for designing challenging classes and providing so much thoughtful
feedback. They are both committed teachers who care deeply about their students. Put simply,
they made me think about the world differently—and I am deeply indebted to them for it.
I would also like to thank the History Department at Haverford College, where I spent a
year as a visiting professor, for funding travel to conduct archival research in the United
Kingdom for this thesis. I would also like to thank everybody, especially Kay Walter, at the
Center for Digital Research in the Humanities at the University of Nebraska-Lincoln (where I am
currently as postdoctoral fellow) for providing such a wonderful intellectual home. Finally, I
want to thank my wife Christina Copland for her love and support.
xii
List of Abbreviations
ANOM Archives Nationales d'Outre-Mer (Aix-en-Provence, France)
BNF Bibliothèque Nationale de France (Paris, France)
BL The British Library (London, UK)
CO Colonial Office
DIV Division
DFC Dépôt des Fortifications des Colonies
FM Fonds Ministériel
GIS Geographic Information System
HGIS Historical Geographic Information System
K.Top King’s Topographical Collection
LAC Library and Archives of Canada, Ottawa, Canada
n.d. not dated
P Pièce
PF Portfeuille
PRDH Programme de recherche en démographie historique (Université de Montréal)
RAPQ Rapport de l'Archiviste de la Province de Québec ([Québec] : Ls-A. Proulx, 1921-
1961) 40 vols
SH Service Hydrographique
SSI Spatial Sciences Institute
TNA The National Archives (Kew, UK)
USC University of Southern California
WO War Office
xiii
Abstract
This project employs a mix of archival and digital techniques to evaluate an extensive
geographic survey of the St. Lawrence River valley, conducted under auspices of the Quebec
Governor James Murray following the 1760 British conquest of Canada. I show how scholars
have misread the historical evidence surrounding the production and dissemination of this survey
by ignoring the patronage motives of those involved in its production. I also use archival
evidence to examine contemporary practices and ideas surrounding maps and accuracy. I discuss
how I built a Historical Geographic Information System by manually creating and spatially
adjusting vector data from Murray’s personal copy of the survey (which was composed of forty-
four individual map sheets measuring forty-five by thirty-six feet when fully assembled). This
dataset allowed me to establish that the survey demonstrated a high degree of spatial accuracy
for the eighteenth century. Here I also discuss methods for retroactively creating administrative
boundaries, which were undocumented in the period. This allowed me to create spatial
interoperability with other contemporary quantitative records—historical censuses and parish
registers, which recorded births, marriages, and deaths—to evaluate the reliability of these
various administrative technologies of state during the early modern period. I conclude that while
the historical censuses undercounted people, the survey and parish registers support each other’s
conclusions, which suggest their demographic accuracy. This work serves as a proof of concept
for a much larger spatial humanities project that would employ these same techniques to digitally
process the series of other geographic surveys conducted throughout British North America
between 1765 and 1777 to capture a geographic snapshot of the late colonial period.
1
Chapter 1: Introduction
. . . In that Empire, the Art of Cartography attained such Perfection that the map
of a single Province occupied the entirety of a City, and the map of the Empire,
the entirety of a Province. In time, those Unconscionable Maps no longer
satisfied, and the Cartographers Guilds struck a Map of the Empire whose size
was that of the Empire, and which coincided point for point with it. The following
Generations, who were not so fond of the Study of Cartography as their Forebears
had been, saw that that vast map was Useless, and not without some Pitilessness
was it, that they delivered it up to the Inclemencies of Sun and Winters. In the
Deserts of the West, still today, there are Tattered Ruins of that Map, inhabited by
Animals and Beggars; in all the Land there is no other Relic of the Disciplines of
Geography. Purportedly from Suárez Miranda, Travels of Prudent Men, Book
Four, Ch. XLV, Lérida, 1658
-Jorge Luis Borges, “On Exactitude in Science”
1
Following the British conquest of Canada in 1760, General Jeffrey Amherst, the commander-in-
chief of the British Army in North America, instructed James Murray, the military governor of
Quebec,
2
to have a survey made of the newly conquered territory. Murray complied, sending
seven teams of military engineers into the field. The surveying occurred through the early
months of 1761. Then draftsmen visually translated the measurements of the surveyors into at
least seven sets of maps, a process that continued until the end of 1762. Figure 1 shows how
these mapmakers synthesized this geographic knowledge onto a single sheet of paper. The most
elaborate copies measured forty-five by thirty-six feet when the forty-four individual sheets were
1
In Jorge Luis Borges: Collected Fictions, translated by Andrew Hurley (New York, N.Y.: Penguin Books, 1998),
325.
2
I have generally chosen to not include diacritical marks for Canadian place names in this thesis because in the main
period of this study, the British did not generally use them. However, I do reproduce them for the names of people
and organizations, the titles of scholarly publications, place names for scholarly publications (i.e. the place name as
it is given by the publication), and when I am quoting original archival material.
2
assembled together. Figure 2 illustrates the forty-four sheets laid over the modern landscape as
revealed by aerial photography.
3
Figure 1: “PLAN of the Settled Part of the Province of Quebec Reduced from the large Survey to
serve as an Index Plan To the Large Sheets” (c.1762), LAC, NMC135035, (henceforth referred
to as the composite map)
3
For the scholars discussing this geographical survey of the St. Lawrence River, see Nathaniel Shipton, “General
James Murray’s Map of the St Lawrence” Cartographica 4:2 (December 1967): 93–101; Douglas W. Marshall,
“The British Military Engineers 1741-1783: A Study of Organization, Social Origin, and Cartography” Dissertation
(University of Michigan, 1976); James Shields, “The Murray Map Cartographically Considered: A Study of General
James Murray’s Survey of the St. Lawrence River Valley in 1761” Dissertation (Queen’s University, 1980); Stephen
Hornsby, Surveyors of Empire: Samuel Holland, J. W. F. Des Barres, and the Making of the Atlantic Neptune
(Montreal: Mcgill Queens University Press, 2011); Alexander James Cook Johnson, “Charting the Imperial Will
Colonial Administration & the General Survey of British North America 1764-1775” (Dissertation: University of
Exeter, 2011); S. Max Edelson, The New Map of Empire: How Britain Imagined America Before Independence
(Cambridge, Massachusetts: Harvard University Press, 2017).
3
Figure 2: Study Area. St. Lawrence River Valley with the Superimposed Sheets of the Murray
Atlas (c.1762), LAC, NMC13036-13078. Present-day aerial photography sources: Esri,
DigitalGlobe, GeoEye, i-cubed, USDA, USGS, AEX, Getmapping, Aerogrid, IGN, IGP,
swisstopo, and the GIS User Community.
In a nutshell, this was the most detailed and expansive survey of North America
completed to that date. The most comprehensive versions of the atlas depicted roads, walls,
dwellings, churches, mills, other buildings, place names (also known as toponyms), waterways,
navigational information, the military positions of British and French armies during the 1759-60
campaigns, and terrain that included forest, meadow, cultivated agricultural land, marsh, swamp,
and various combinations thereof. Additionally, the draftsmen recorded notes about each parish,
usually a combination of history and geographic description with the results of a 1760 census
(undertaken independently from the geographic survey). The level of detail and the scope of the
project were at the same time grandiose and an overload of information. Indeed, this ambitious
masterpiece, henceforth referred to as the Murray Atlas, could have inspired the absurdist short
4
story of Jorge Luis Borges “On Exactitude in Science” in which the tattered remains of map
made at the scale of 1:1 offered refuge to “Animals and Beggars” in the deserts of a once
geography-obsessed civilization.
4
The advent of computing has rendered the absurdity of that short story somewhat moot.
As hard drives and screens have replaced paper as the repository for geographic information, the
scale of 1:1 no longer requires us to shroud the real territory with paper in the manner Borges
described. Indeed, computing and a diverse array of analysis and visualization tools can also
breathe new life into such obsessive data. These new tools can elucidate the patterns and trends
that humans cannot grasp with the naked eye. In other words, while the act of zealously setting
down details might have once obscured, now it adds evidentiary fuel for the analytical fire. Yet
in thinking about the Murray Atlas, we might look to Borges (and a generation of historians of
cartography) for a reminder that we should suspect the rhetorical exercise of “exactitude.” And
here in a nutshell is the point of this project. The Murray Atlas appears impressive but digital
tools can now help us judge (and potentially improve) the quality of the data contained in the
maps. And we can also use digital tools to analyze that data, i.e. to transform the tattered,
weather-beaten, and “Useless” artifact of imperial obsession into something meaningful for
understanding the past.
This thesis develops a conceptual and a methodological framework for translating early
maps—by which I mean those predating aerial photography—into historical geospatial data that
we can be confident are fit for use in undertaking historical geographical analysis. I begin with
some of the vocabulary of Geographic Information Science concerning different elements of
4
Scholars refer to the results of the survey as both the Murray Atlas and the Murray Map so that both titles show up
in citations. Here, I stick to “Murray Atlas” for the sake of consistency.
5
spatial data quality, particularly completeness and positional accuracy.
5
Said differently, how
well did the surveyors and draftsmen record the existence of (completeness) and spatial
relationships between (positional accuracy) the geographic phenomena they depicted? And if we
find error in either of these realms, can we reduce it enough in digitally processing the historical
map to be confident in the quality of the historical geospatial data?
Crucial to this process was archival research. Here I studied the creation and
dissemination of the Murray Atlas to understand what Geographic Information Science refers to
as its spatial data lineage.
6
Questions I asked included: What motivated its production? How
might those motivations have shaped the quality of the data? For what purposes did historical
figures hope to use the map? Was the data quality, in their minds, fit for those uses? How did
those surveying, drafting, or coordinating cartographic labor conceptualize accuracy?
Next, I digitally processed the Murray Atlas within a GIS, meaning that I manually
created vector data on top of digital scans of the historical maps and then spatially adjusted that
vector data to align with our present-day understanding of the terrain. I employed spatial and
quantitative tools to evaluate the positional accuracy of the data represented within the original
historical map. Then I worked with historical censuses and parish registers, two demographic
sources that helped me to evaluate the completeness of dwellings—which are, from a historical
geographic perspective, one of the most important features depicted in the Murray Atlas for their
ability to tell us about human settlement patterns.
I organize this thesis into four parts. The next chapter reviews the relevant literature on
the historical geography of Canada and digital approaches to the history of cartography. I explore
the benefits of bringing these fields more closely together and explain the necessity of grounding
5
For a full overview of these elements, see Pepijn van Oort, “Spatial Data Quality: From Description to
Application” Dissertation (Wageningen Universiteit, 2006), 11-20.
6
van Oort, “Spatial Data Quality,” 14.
6
such endeavors in archival research. Next, in examining scholarship on the Murray Atlas itself, I
argue earlier scholars have fundamentally misread the archival sources, failing to identify the
primacy of patronage motives. In the third chapter, I reinterpret this archival record to show how
these motives shaped the rhetoric of accuracy in discourses about the map. I also situate the
Murray Atlas in a century-long, mutually constitutive relationship between cartography and
human geography in Canada. In Chapter Four, I outline my methods for digitally processing the
Murray Atlas. I discuss spatial distortion within the Murray Atlas, concluding that the surveyors
did a good job given their goals and limitations. I also explain how I worked with the digitized
Murray Atlas to spatially align it with other quantitative sources. In the final chapter, I discuss
these other quantitative sources—five historical censuses taken between 1760 and 1765 by the
British and parish registers of vital events (births, deaths, and marriages) stretching back to the
colony’s founding under the French. I conclude that the Murray Atlas and the parish registers
support each other’s demographic conclusions while the censuses appear to undercount people in
erratic ways.
7
Chapter 2: Towards a Historiographical Synthesis of Archival
and Digital Methods
In October 1690 a fleet of New England vessels sailed up the St. Lawrence River, bombarded the
French city of Quebec, and ultimately retreated. Nearly two centuries later, the historian Francis
Parkman devoted two chapters to this event in his Count Frontenac and New France under Louis
XIV (1877), concluding with a curious footnote:
“In the Bibliothèque Nationale is the original draft of a remarkable map, by the engineer
[Robert de] Villeneuve, of which a fac-simile is before me. It represents in detail the
town and fortifications of Quebec, the surrounding country, and the positions of the
English fleet and land forces.”
7
Parkman relied on this “fac-simile” to describe Quebec’s topography and the location of its
palisades, batteries, and artillery—but he did not know the full story of the map: While
Villeneuve did meticulously survey Quebec and its environs during the mid-1680s, he had
departed the colony before the city was fortified or assaulted.
8
The siege map, which Villeneuve
drafted with a great deal of imagination, was poor evidence for describing the military matchup
in 1690.
This cartographic deception would not merit our attention except that Parkman
bequeathed his facsimile to Harvard College whose librarians recently digitized it, making it
freely available online. They also georeferenced it so that scholars working with a Geographic
Information System (GIS) could import its data, described as “roads, drainage, selected buildings
with names of landowners, fortification, English ships and lines-of-fire, ground cover, and
more.” The logic was that “Historic paper maps can provide an excellent view of the changes
7
Parkman, Count Frontenac and New France under Louis XIV (Boston: Little, Brown, & Company, 1877), 285. For
how Parkman acquired the map, see Parkman to Pierre Margry (Boston: 16 Oct 1875) in Wilbur R. Jacobs, ed.,
Letters of Francis Parkman, 2 vols (Norman: University of Oklahoma Press, 1960), II:85-87.
8
Jean Bochart de Champigny to Jean-Baptiste Antoine Colbert, Marquis de Seignelay (16 Nov 1689) ANOM, FM,
C11A, 10:244-250.
8
that have occurred in the cultural and physical landscape.” Of course this only is true if we can
trust the historic paper map and it is indeed the authentic original—neither condition being true
in this case. The conundrum is that few historians have the deep technological knowledge
necessary to create, manage, or analyze spatial data while few technicians have the archival skills
to interrogate the originality, authenticity, or truthfulness of historical sources.
9
2.1. The Underuse of Historical Maps in the Spatial Humanities
Within the past two decades a handful of humanists, most without formal geographic
training, have begun integrating geospatial technology into their scholarship. This field of
“spatial humanities” emerged out of many disparate, sometimes epistemologically irreconcilable
intellectual trends. It draws inspiration from the circulation of people, goods, and ideas described
by historical geographers. It engages the quantitative methods adopted and developed by social
historians of the 1960s and 70s who believed that numbers could help give voice to historically
disempowered peoples.
10
Yet it was also deeply influenced by poststructuralist theoreticians who
had conceptualized the subjectivity of space while rejecting the positivist impulse undergirding
quantitative social sciences.
11
Surprisingly, scholars working in the spatial humanities have not
engaged deeply with the history of cartography, which emerged in the 1980s as historians, art
9
Harvard College Library, “Quebec, Canada, ca. 1690 (Raster Image)” in the Harvard Map Collection Digital Maps
(accessed 7 September 2014) available: http://vc.lib.harvard.edu/vc/deliver/~maps/G3454_Q4R4_1690_V5_1855.
10
On quantitative history, see J. Dennis Willigan and Katherine A. Lynch, Sources and Methods of Historical
Demography (New York, N.Y: Academic Press, 1982); and especially the historiographical overview in Pat
Hudson, History by Numbers an Introduction to Quantitative Approaches (London: Arnold, 2000).
11
On Poststructuralism, see Jurgen Habermas, The Structural Transformation of the Public Sphere: An Inquiry into
a Category of Bourgeois Society Translated by Thomas Burger (Cambridge, MA: MIT Press, 1962); Yi-Fu Tuan,
Space and Place: The Perspective of Experience (Minneapolis: University of Minnesota Press, 1977); Michel de
Certeau, The Practice of Everyday Life (Berkeley, CA: University of California Press, 1988); Michel Foucault,
Discipline and Punish: The Birth of the Prison 2nd Vintage Books ed. (New York: Vintage Books, 1995).
9
historians, literary scholars, and geographers collaborated to understand the significance of maps
to historical development.
12
Historical GIS (HGIS) projects have explored a variety of geographic scales ranging from
analysis of global networks and the dissemination of print matter within Europe to the changing
environments of cities and towns. The small-scale projects rarely rely directly on historical maps
but work with digital gazetteers (which themselves may have been constructed with the aid of
historical maps) providing the locations of non-extant toponyms.
13
The large-scale projects often
use georeferenced historical maps to integrate quantitative and textual data derived from archival
sources.
14
Usually this is done for aesthetic reasons rather than analytical ones. Rarely do these
projects treat the historical maps as sources of valuable geospatial data in of themselves. Most of
this work concerns the nineteenth and twentieth-centuries, which reflects the broader field of
quantitative historical scholarship and limitations existing within the earlier documentary record.
Moreover the small number of historically oriented geospatial datasets (especially before 1800)
has limited the growth of this potentially dynamic field. Its development depends on scholars
with dual areas of expertise who are able to both process and evaluate historical sources to create
new datasets.
15
12
For the most influential work on American Historical Geography, D.W. Meinig, The Shaping of America: A
Geographical Perspective on 500 Years of History (New Haven: Yale University Press, 1986). For influential works
on the history of cartography see J. B. Harley and David Woodward, eds., The History of Cartography. 6 vols
(Chicago: University of Chicago Press, 1987); Harley, “Silences and Secrecy: The Hidden Agenda of Cartography
in Early Modern Europe” Imago Mundi (January 1988), 40: 57–76.
13
Fiona Black, Malcolm Black, and Bertrum MacDonald, “Geographic Information Systems: A New Research
Method for Book History” Book History (1998), 1: 11–31; Franco Moretti, Atlas of the European Novel, 1800-1900
(London: Verso, 1998) and Graphs, Maps, Trees: Abstract Models for Literary History (London: Verso, 2007); Ana
Crespo Solana, ed., Spatio-Temporal Narratives: Historical GIS and the Study of Global Trading Networks (1500-
1800) (Newcastle upon Tyne: Cambridge Scholars Publishing, 2014);
14
Loren J. Siebert, “Using GIS to Document, Visualize, and Interpret Tokyo’s Spatial History” Social Science
History (2000), 24: 537–74; Brian Donahue, The Great Meadow: Farmers and the Land in Colonial Concord (New
Haven: Yale University Press, 2004).
15
For edited books that include projects at a mix of different scales see, Anne Kelly Knowles, ed., Placing History:
How Maps, Spatial Data, and GIS Are Changing Historical Scholarship (Redlands, CA: ESRI Press, 2008);
Jennifer Bonnel and Marcel Fortin, eds., Historical GIS Research in Canada (Calgary: University of Calgary Press,
10
This project, then, largely emerges from a belief that to date historical maps have been
underutilized and misused as a source base for recreating and analyzing historical landscapes,
especially early modern ones. Historians of cartography have been interested in positional
accuracy for the past half-century, originally using it as a kind of barometer for the scientific
progress of civilization.
16
In the 1980s, scholars in this field began distancing themselves from
this notion as they embraced the poststructuralist argument that maps are subjective documents
shaped by the worldviews of the society producing them. The question of accuracy has
reappeared more recently as a handful of historians of the map have integrated geospatial tools
into their analytical toolkit. They have analyzed positional accuracy to infer how mapmakers
compiled and processed their information.
17
Yet the assumption of extreme cartographic
subjectivity raises doubts about the completeness of what has been depicted and, therefore,
undermines the premise of using historical maps for historical geography
This all begins to explain why historians of cartography have rarely attempted to bridge
their work with that of the historical geographer. We might examine one prominent article that
does do this work for more insight into its difficulty. In 2010, Levin, Kark, and Galilee used 375
historical maps to understand how nomadic peoples in Palestine’s Negev Desert became
sedentary across the nineteenth and early twentieth centuries.
18
For them working on a frequently
mapped area, surveyed by many different regimes, and employing a mix of techniques including
2014). For overview of the spatial humanities more generally, see Ian Gregory and Paul S Ell, Historical GIS:
Technologies, Methodologies, and Scholarship (Cambridge: Cambridge University Press, 2007); David J,
Bodenhamer, John Corrigan, and Trevor M Harris, The Spatial Humanities: GIS and the Future of Humanities
Scholarship (Bloomington: Indiana University Press, 2010).
16
W. R. Tobler, “Medieval Distortions: The Projections of Ancient Maps.” Annals of the Association of American
Geographers 56, no. 2 (June 1966): 351–60; Shields, “The Murray Map Cartographically Considered”
17
Brooks Pearson, “Comparative Accuracy in Four Civil War Maps of the Shenandoah Valley: A GIS Analysis” The
Professional Geographer 57, no. 3 (2005): 376–94; Richard J. A. Talbert and Tom Elliott, “New Windows on the
Peutinger Map of the Roman World” in Placing History, ed. Knowles, 199-218; Christopher D. Lloyd and Keith D.
Lilley, “Cartographic Veracity in Medieval Mapping: Analyzing Geographical Variation in the Gough Map of Great
Britain” Annals of the Association of American Geographers 99:1 (January 2009): 27–48.
18
Noam Levin, Ruth Kark, and Emir Galilee, “Maps and the Settlement of Southern Palestine, 1799-1948: An
Historical/GIS Analysis” Journal of Historical Geography 36 (2010): 1–18.
11
aerial photography gave them a confidence in the completeness of geospatial data that they
extracted. Their discussion encompassed more than the geographic phenomena depicted within
the historical maps but also considered agency of those maps as objects (e.g. in shaping and
legitimizing landscapes as opposed to simply “reflecting” them) along with the imperialism
underlying their creation (which also exerted agency in shaping landscapes and legitimizing or
delegitimizing through representation the people within those landscapes).
The time, energy, and expense involved (which they discuss) in producing such thorough,
thoughtful scholarship probably explain why we don’t see more. Their work points to a second
dilemma, namely that the phenomena depicted within the historical maps shape, to some extent,
the kinds of questions it is possible to ask within a GIS environment—which perhaps limits the
possibility for a single research agenda or consistently replicable method of spatial analysis.
Finally, their work points to a third difficulty for earlier maps—i.e. those before their study
period—namely that without aerial photography it is difficult to assess completeness in an
evaluation of accuracy.
Some historical geographers of Canada have used historical maps as evidence to grasp at
some of the qualities shaping settlement patterns—but much of this work was done before the
advent of GIS or statistical software. Courville who has authored the most recent and definitive
account of the historical geography shaping the present-day region of Quebec mostly synthesized
older scholarship (much of it his own) from the 1960s and 1970s. Courville and his
contemporaries, for example, may have tallied the number of seignieurial grants to different
classes of people (nobles, clergy, merchants, etc.) but did not have had the tools to push their
inquiry much further. Did any kind of spatial logic govern the distribution of those grants? Did
12
merchants do a better job peopling their grants than clergy? Courville was silent on the Murray
Atlas, surprising given his extensive use of earlier French and later British maps.
19
2.2 Scholarship on the Murray Atlas
This chapter began with a quote from the Harvard Libraries explaining their rationale for
digitizing and georeferencing historical maps. That rationale could apply to the Murray Atlas as
well. Before digitally processing it, we must first do our due diligence so that we do not repeat
their mistake. We must evaluate the reliability of the Murray Atlas. Have scholars raised any red
flags? Or do we find records in the archives that undermine it?
Scholarship on the Murray Atlas, itself, has focused on the context of its production. This
began in 1967 with Nathaniel Shipton’s brief article inspired by the copy held by the William L.
Clements Library where Shipton served as the Map and Print Librarian. After reviewing the
chronology of the survey and drafting, Shipton hypothesized about whom had received which of
the surviving copies (although he was stumped when it came to determining whom had
originally possessed Clements Library’s copy). Douglass Marshall followed a decade later with a
dissertation on British military engineers in North America, devoting special attention to the
Murray Atlas as the only one of “five military mapping projects . . . in North America between
1761 and 1775” that was sponsored exclusively by the army. Marshall fleshed out Shipton’s
timeline, the biographies of the military engineers and draftsmen involved, and situated the
Canadian survey within the mapping practices that had originated from a 1740s survey of
Scotland.
20
19
Serge Courville, Quebec: A Historical Geography translated by Richard Howard (Vancouver: UBC Press, 2008),
originally Le Territoire: Atlas Historique du Québec (Quebec: Presses de l'Université Laval, 1997).
20
Shipton, “General James Murray’s Map”; Marshall “The British Military Engineers.”
13
The fullest treatment of the Murray Atlas came in a 1980 geography dissertation by
James Shields. Shields began with a wide-ranging literature review, which included
historiographical discussion of the British occupation of Quebec and contemporary surveying
techniques. His second chapter narrated the military conflict between the British and French in
North America, through the British conquest of Canada. Next, Shields canvassed earlier mapping
projects in the St. Lawrence River Valley, quickly moving through the French period and
offering a more sustained examination of British reconnaissance mapping. This led into a
sustained discussion of the Murray Atlas during chapter Four, mostly distinguished from the
earlier scholarship by its greater depth rather than the novelty of its conclusions. Most
significant, Shields located the intense intrapersonal conflict—noted but not explained by
Shipton and Marshall—between the mapmakers under Murray in “the rivalry of engineers and
battalion officers.”
21
Breaking the most new ground was chapter Five in which Shields offered his
“cartographic assessment” of the Murray Atlas. First, Shields revisited Shipton’s assessment of
who received which copies. Next he performed a visual and stylistic analysis of those different
copies, noting that “the presentation of landforms . . . lacks a consistency and a degree of
accuracy.”
22
Mostly, he attributed these inconsistencies to the stylistic tics of the draftsmen—and
as such situates the Murray Atlas within the uneven process by which British military
cartography was codified.
Shields took special issue with the “method of portraying farmland” for not bearing any
“resemblance to the true nature of the landholdings in the St. Lawrence River valley; that of long
21
Shields, “The Murray Map Cartographically Considered,” 91.
22
Shields, “The Murray Map Cartographically Considered,” 124.
14
narrow lots stretching away from the river.”
23
He also observed certain inconsistencies in the
depictions of the three urban areas as compared against more detailed town plans, explaining that
the very existence of those plans granted the surveyors some measure of cartographic license in
depicting the towns at a smaller scale. He praised “the concentrated efforts of the surveyors” in
the countryside.
24
In concluding this visual analysis, Shields argues that controlling the recently
conquered population was the rationale of the mapping project. As he explains, the “hierarchy in
their choice of features to be portrayed” reflected the “concern of the British for effective control
of the community.”
25
Finally, Shields applied the—what was at the time “new”—practice of cartometry to
quantify whether the Murray Atlas could “fulfill the basic requirement of portraying the
landscape with an accuracy which could be depended upon for military purposes.”
26
He
accomplished this—without the assistance of computers—by selecting “a number of features . . .
for locational comparison with identical points on modern maps . . . [in] the regions immediately
surrounding Quebec City, Trois Rivieres and Montreal.”
27
This analysis should be useful for us in thinking about using the map for a spatial
humanities project but it is undermined by a questionable methodology. Shields claims that for
these physical and human features to have qualified as his control points, they had to have
“existed in the same position since the eighteenth century.”
28
A close examination of his control
points, which can be seen in Figure 3, calls this assertion into question: It is in no way clear that
the points he chose actually persisted on the landscape. We might take the Beauport area as one
23
Shields, “The Murray Map Cartographically Considered,” 129.
24
Shields, “The Murray Map Cartographically Considered,” 132.
25
Shields, “The Murray Map Cartographically Considered,” 134.
26
Shields, “Murray Map Cartographically Considered,” 135.
27
Shields, “Murray Map Cartographically Considered,” 137.
28
Shields, “Murray Map Cartographically Considered,” 138.
15
example, it being one of the places he identified as having the most linear error at 66.4%. The
black circles represented the points he used in his nearest neighbor measure and seem to
correspond to street intersection in the Murray Atlas—
Figure 4 provides the view from the Murray Atlas of the same area depicted in Figure 3.
However, we do not know that there is indeed this kind of continuity in where roads lay or met.
The Murray Atlas does not generally name streets (though there is in one case a somewhat
ambiguous toponym that might refer either to a street or a parish) so these are unsubstantiated
assumptions that Shields brings to his reading of the map.
Figure 3: Detail of Beauport from Shields, “Cartometric Assessment of the Quebec City Region”
in “Murray Map Cartographically Considered,” 150.
16
Figure 4: Detail of Beauport in the Murray Atlas. Are the roads depicted here the same ones as
existed when Shields was writing in 1980? Given changes in most urban landscapes since 1760,
it would be surprising if they were.
After identifying control points and measuring distances between them (translating them
according the map scale), he compared the results from Murray’s personal copy of the Murray
Atlas and (what were then) present-day topographic maps to obtain the linear error. He did this
across both longer and shorter distances, concluding that “the map, while reasonably accurate
overall, had some areas of localized error far in excess of the mean.”
29
He attributed this
localized error to the mistakes of particular surveyors and the haste in which the survey was
taken. In my fourth chapter (section 0) I will present a very different view of positional accuracy
in the Murray Atlas. I would suggest now that his questionable choice of control points explains
why Shields found such a high rate of localized error in the places he did.
29
Shields, “Murray Map Cartographically Considered,” 139.
17
Ultimately, Shields judges the Murray Atlas was “an improvement on earlier maps of the
region, but that for reasons of time constraints and manpower problems, the attainment of high
accuracy was unfulfilled.”
30
Yet
this hedge does not answer the original question, namely
whether the survey was accurate enough to suit the military purposes for which it was intended.
And this is indeed a question worth our revisiting—and more important, asking the inverse, i.e.
What might the Murray Atlas’ degree of accuracy tell us about why it was made?
Recent scholars have glossed over Shield’s arguments about error in the Murray Atlas,
instead recommending it as one of the most advanced, precise cartographic undertakings to that
date.
31
They also focus on the Murray Atlas only briefly. This is more or less a narrative strategy
on their parts, making the survey a prelude to a larger cartographic moment in the late
eighteenth-century Atlantic World. This new scholarship, then, has mostly synthesized selective
parts of the earlier work without revisiting the evidence or interpretations. The exception is
Edelson who has consulted a report on the government of Quebec drafted by Murray, using it as
the primary lens for understanding the mapping project. While this is a reasonable enough
instinct, I show in the following chapter (section 0) that Murray did not intend text and map to
work together in quite the way that Edelson assumes. Most problematic, this group of scholars
has largely failed in doing their due diligence. They have largely ignored Shield’s most
important chapter “Cartographic Assessment of the Murray Map” and have instead readily
accepted the eighteenth-century rhetoric of accuracy deployed in the map and the discourses that
surrounded it. In the next chapter I dive into the archival sources to explain why this is so
problematic.
30
Shields, “Murray Map Cartographically Considered,” 152.
31
These include: Johnson, “Charting the Imperial Will”; Hornsby, Surveyors of Empire; Edelson, The New Map of
Empire.
18
Chapter 3: Historical Overview of Cartography in Canada during the French
Regime and through the British Conquest
This chapter recounts the history of cartography in Canada, showing how the aims of mapmakers
and their ways of cartographic thinking helped to shape on-the-ground settlement patterns. Next,
we consider the continuity of cartographic styles and practices across the French and British
regimes. This helps us situate contemporary ideas and definitions of accuracy within the
collection, representation, and use of what we today refer to as geospatial data. Finally, this
chapter revisits the specific archival sources surrounding the creation and dissemination of the
Murray Atlas.
3.1. The French Mapping of Canada and the Shaping of the Human Geography
The French mapping of America commenced in 1524 when King François I
commissioned the Florentine navigator Giovanni da Verrazano to sail across the Atlantic. Most
of the maps from this voyage, which ultimately took Verrazano and his crew along the American
coast from Florida to Cape Breton, have been lost.
32
Other French expeditions in the sixteenth
century had produced only simple coastal surveys. More significantly for the history of
cartography, mapmakers and printers in France incorporated the textual accounts of these
explorers into their work. The 1547 Vallard Atlas, for example, drew heavily upon the work of
the explorer Jacques Cartier in its depiction of Canada. Maps drafted for royal eyes could be
incredibly ornate, filled with notes and icons meant to pique royal interest in funding overseas
32
One relatively simple sea chart by Verrazzano’s brother survives at the Vatican Library where it is currently on
display. For a reproduction, see Edward Luther Stevenson, Maps Illustrating Early Discovery and Exploration in
America, 1502-1530 (New Brunswick, NJ: 1906); William F. E. Morley, “Verrazzano, Giovanni da,” in Dictionary
of Canadian Biography, vol. 1 (1966, revised 1979).
19
ventures. Northeastern America did not figure prominently into these early fantasies, however,
and initial expeditions to Canada resulted in little more than a French fishing outpost.
33
Significant French interest in this region only developed around 1600 when felt tipped
hats lined with the pelts from beavers, abundant in northeastern America, became fashionable in
Europe. During the summer months French merchants sailed to the Canadian Maritimes to
acquire pelts from Native trading partners. Initially pelts resold in Europe for roughly two
hundred times their original cost but as more ships made the transatlantic voyage, discerning
Natives raised their prices.
34
The French King Henry IV granted a trade monopoly to reduce
competition, requiring the monopoly holder to establish a permanent settlement in Canada. The
cartographer Samuel de Champlain participated in these early seventeenth-century corporate
expeditions, producing maps created from his own observations as well as from Native
knowledge meant to survey natural resources, facilitate subsequent navigation, locate ideal sites
for a French settlement, establish French territorial claims, and attract settlers along with capital
for the nascent French colony.
35
When initial attempts to establish a colony along the Acadian coast failed, the
geographically savvy Champlain recommended building a post where the north and south shores
of St. Lawrence River began converging, just upstream of the Isle d’Orleans—a strategic node
that (if properly defended) would control access to the northern continental interior. The colony
33
For an example of ornate French cartography based on the textual accounts of explorers, see Pierre Desceliers’
considerably large (roughly 40 square feet) map of the world drawn on vellum in 1550 for either King Henri II of
France or the Duc de Montmorency. At the British Library, Additional Manuscript 24065; Vallard Atlas, HM 29,
[fol. 9] (Canada), Huntington Library; Frank Lestringant and Monique Pelletier, “Maps and Descriptions of the
World in Sixteenth-Century France” in Harley and Woodward, History of Cartography, III:1463-1479; Peter C.
Mancall, “The Raw and the Cold: Five English Sailors in Sixteenth-Century Nunavut” The William and Mary
Quarterly 70:1 (January 1, 2013): 3–40, see 21-22 for discussion of the Vallard Atlas.
34
J. F. Crean, “Hats and the Fur Trade” The Canadian Journal of Economics and Political Science / Revue
Canadienne d’Economique et de Science Politique 28:3 (August, 1962): 373–86;
35
W. J. Eccles, The Canadian Frontier, 1534-1760 (New York: Holt, Rinehart and Winston, 1969), 12-34; Conrad
E. Heidenreich, “The Mapping of Samuel de Champlain, 1603–1635” in Harley and Woodward, History of
Cartography, III:1538-1549; David Hackett Fischer, Champlain’s Dream (New York: Simon & Schuster, 2008).
20
during these early years consisted of a trade outpost around present-day Quebec City, destroyed
by English privateers in 1629. Champlain drafted a map showing French claims for the
negotiations in London. This non-extant manuscript map likely informed his 1632 engraved map
(accompanying his pamphlet Les voyages de la Novvelle France) aimed at attracting people and
capital to resettle Canada, portrayed a land of abundance filled with trees, agricultural land,
botanical specimens, and animals with valuable pelts.
36
In reestablishing the colony, the Compagnie de la Nouvelle France hoped to do a better
job peopling and protecting it. To these ends, they sent the Jesuit-educated engineer and
mapmaker Jean Bourdon to help reestablish the colony. One challenge was militaristic. During
these years the fur trade had cemented alliances between the French and various Native groups,
particularly the Hurons and Algonkins, but these alliances also drew the French into prolonged
warfare with the Iroquois League.
As attacks intensified in the 1640s, Bourdon travelled to France in search of aid,
furnishing two unembellished maps to secure goodwill. One of these can be seen below in Figure
5, which depicts Quebec City in the bottom left corner. Aspirational in the sense that they
projected imaginary lines onto the landscape, Bourdon’s maps revealed attempts to institute in
Canada a seigniorial land tenure system in which the Compagnie granted large parcels of land to
seigneurs who in turn were responsible for subdividing, “improving,” and peopling those grants.
When the colony faltered in 1660, Bourdon again journeyed to France. On this occasion he
supplied a small-scale map of Quebec, which can be seen in Figure 6. Titled “True plan of high
36
Samuel de Champlain, Les voyages de la Novvelle France occidentale, dicte Canada . . . (Paris: Clavde Collet,
1632); Heidenreich, “The Mapping of Samuel de Champlain”; Tim McNeese, The St. Lawrence River (Philadelphia:
Chelsea House Publishers, 2005), 2-6; James H. Thorp, Gary A. Lamberti, and Andrew F. Casper, “St. Lawrence
River Basin” in Arthur C. Benke and Colbert E. Cushing, eds., Rivers of North America (Boston: Academic Press,
2005), 983-1030; Raymonde Litalien, Denis Vaugeois, and Jean-François Palomino, Mapping a Continent:
Historical Atlas of North America, 1492-1814 (Québec: Les éditions du Septentrion, 2007), 83-88; Fischer,
Champlain’s Dream.
21
and low Quebec,” this map similarly projected imaginary lines to plan city development. Mostly,
however, the Compagnie did not invest much into realizing these plans, leaving the French
settlers to their own devices.
Figure 5: Jean Bourdon “Carte depuis Kébec jusque au cap de Tourmente” (1641), BNF, GED-
8070(RES).
22
Figure 6: Jean Bourdon, “Vray Plan du haut & bas de quebec comme il est en l[’]an 1660”
(c.1660), ANOM, DFC 341C.
This began to change, however, once Louis XIV revoked the charter of the bankrupt
Compagnie de la Nouvelle France in 1663, proclaiming Canada a royal colony. Throughout this
decade, the French state sponsored a wave of male and female immigrants to Canada, hoping to
create a demographically and economically self-sufficient colony that could contribute to state
coffers rather than drain metropolitan resources. Population growth after this point (under French
rule) hinged increasingly upon natural increase as the number of transatlantic immigrants
23
dwindled. Imperial reorganization coincided with the invigoration of state cartography, owing to
Finance Minister Jean-Baptiste Colbert who upon his appointment in 1661 implemented
domestically a state culture of intelligence gathering that privileged maps for reporting and
integrating economic, legal, demographic, natural, ecclesiastical, and military information. In
this context, governors and other Canadians began regularly sending maps (usually containing
the decorative hallmarks of patronage objects) to Versailles to advance their personal agendas. In
other words, maps in this transatlantic bureaucratic context served a blend of patronage and
evidentiary purposes. The appearance of accuracy often mattered more than actual accuracy—
who would know the difference, after all?
37
By the 1680s, the French state began sending military engineers and other metropolitan
mapmakers as part of an increasingly mercantilist colonial agenda. Two partial surveys of
Canada were made under Louis XIV during 1685−6 and 1709. Examples from the earlier survey
of Robert de Villeneuve can be seen in Figure 7 (Quebec City) and Figure 8 (the environs around
Quebec City). The French governor sent these to demonstrate the success of the centralized state
planning in shaping concentrated settlement patterns into the terrestrial interior. An example
from the second survey, depicting the lots of land along St. Lawrence River can bee seen in
Figure 9. Figure 10 shows a detail from Figure 9 of the city of Trois Rivieres with a better view
of these lots and the textual descriptions naming the people to whom those lots belonged. The
archival context of this survey is less clear but it would seem these maps were meant to
demonstrate the success of the Seigniorial system of land settlement in which a Lord or other
37
In 1663 Colbert requested that officials throughout France send “accurate and detailed maps of each province and
généralité” and employed the geographer Nicolas Sanson to process the information. At the decade’s end, Colbert
hired the Italian mathematician Jean-Dominique Cassini to conduct a national survey of France. Konvitz,
Cartography in France, 1-8, quote on 2; Jacob Soll, The Information Master: Jean-Baptiste Colbert’s Secret State
Intelligence System (Ann Arbor: University of Michigan Press, 2011), 70-6, 99; Brotton, A History of the World in
12 Maps, 294-336.
24
notable person was granted a large parcel of land and was responsible for subdividing it to attract
settlers.
38
Figure 7: Robert de Villeneuve, “Plan De La Ville Et Chateau de Quebec, Fait En 1685, Mezvrée
Exactement” (1685), ANOM, DFC, 349B.
38
Nicholas Gliserman, “Landscapes of Conflict: Cartography and Empire in Northeastern America, 1680-1713”
(Dissertation, University of Southern California, 2016), 100-157.
25
Figure 8: Villeneuve, “CARTE DES-ENUIRONS DE QVEBEC EN LA NOUVELLE FRANCE
Mezuré Sur le Lieu Tres-Exactement, En 1685, Et 86, PAR LE S
r
DEVILLENEVVE Injenieur
du Roy” (1685-6), BNF,
26
Figure 9: Gédéon de Catalogne and Jean Baptiste de Couagne, “Carte du gouvernement des trois
rivieres” (1709) BNF, SH18, PF127, DIV2, P4
Figure 10: (Detail from Figure 9) The walled city of Trois Rivieres, surrounded by long, narrow
lots.
These two partial surveys suggest the staying power of Bourdon’s templates for peopling
Canada Three major towns (Quebec, Montreal, and Trois Rivieres) closely resembled
northwestern French ports divided between “a commercial lower town and an administrative and
military upper town.” Baroque architecture, geometric garden plots, and religious buildings
framed the upper city while a denser matrix of commercial buildings and residences hugged the
27
river on the lower ground. Between the cities and villages farmers gradually peopled long,
narrow plots fronting the St. Lawrence. Arriving from a variety of regions, these habitants
blended together a variety of landscape vernaculars. Rapids just upstream of Montreal barred the
entry of large sail-based European vessels and limited further settler colonialism. Beyond these
natural limits, soldiers built sparse wooden fortifications to conduct and protect the fur trade,
which dominated the colony’s economy. French missionaries who travelled widely throughout
Indian country lured Christian converts to settle communities in the shadow of French towns.
39
After the death of Louis XIV, however, cartographic exchanges between metropole and
colony changed. On the one hand, the French state invested more resources into building and
planning elsewhere in French North America, sending cartographically talented engineers to
Louisburg, Louisiana, and the Mississippi River. Canadians furnished no more ambitious surveys
of the St. Lawrence River throughout the remainder of the French regime—although they did
send fortification schematics and a few town plans.
40
Immigration declined after the state-
sponsored wave of the 1660s and would not seriously pick up again until the Seven Years War,
limiting the ability of overseas metropolitan officials to easily create enduring templates for
settlement patterns in new places.
41
The village system championed by Colbert during the 1660s
in the environs of Quebec remained in place, for example, but would not be established
elsewhere in the colony.
39
Cole Harris, “Retracing French Landscapes in North America” in Michael P. Conzen, ed., The Making of the
American Landscape. 2nd ed. (New York: Routledge, 2010), 73-90, quote on 79; Serge Courville and Richard
Howard, Quebec: A Historical Geography (Vancouver: University of British Columbia Press, 2008).
40
Gliserman, “Landscapes of Conflict,” 213-261, 416-427.
41
Leslie Choquette, Frenchmen into Peasants: Modernity and Tradition in the Peopling of French Canada
(Cambridge, Mass: Harvard University Press, 1997).
28
3.2. British Mapmaking and the Invasion of Canada during the Seven Years War
Civil War and Revolution in seventeenth-century England hindered the kind of French
centralized cartographic bureaucracy from developing there—even as the late Stuart monarchs
Charles II and James II were envious of and tried implementing it. Somewhat ironically in the
eighteenth century, the state cartography of France originally developed under the auspices of
Colbert and Louis XIV, continued to exert an outsize influence on British cartography: Droves of
Huguenot engineers, trained under the influence of the master engineer of the French state,
Sébastian le Prestre Vauban, defected to Britain. Printed maps similarly traversed the English
Channel in these years. The cumulative result was a common cartographic language and
aesthetic in the British and French empires.
42
While the British military had institutionalized the training of engineers in the 1740s,
surveying and map drafting did not initially figure much into this education. Instead, military
engineers acquired these skills in the field so that throughout the eighteenth century, their
knowledge circulated in more decentralized ways and catered to local rather than metropolitan
agendas. Around the same time as the establishment of this engineering school, the military
expanded the number of copyists working at the Tower of London. Only in 1752 did this
“Drawing Room” become “a recognized department.”
43
An 1841 fire makes it difficult to assess
the significance of the Drawing Room as a cartographic storehouse during the Seven Years War.
In any case, it seems that it sent few maps or draftsmen to North America during the Seven
Years War. British officers frequently noted the difficulty in locating existing maps, as well as
42
Peter Barber, “Necessary and Ornamental: Map Use in England Under the Later Stuarts, 1660-1714.” Eighteenth-
Century Life 14, no. 3 (1990): 1–28; Barber and A. Stuart Mason. “‘Captain Thomas, the French Engineer’: And the
Teaching of Vauban to the English” Proceedings of the Huguenot Society 25 (Autumn 1991): 279–87; Mary
Sponberg Pedley, The Commerce of Cartography: Making and Marketing Maps in Eighteenth-Century France and
England (Chicago: University of Chicago Press, 2005).
43
Marshall, “Military Maps of the Eighteenth-Century and the Tower of London Drawing Room” Imago Mundi 32
(January 1, 1980): 23.
29
copyists, surveyors, and cartographic draftsmen. Surveying equipment was often unavailable—
frequently requested but rarely sent.
44
The Seven Years War saw British mapmakers flood into North America. In 1756, before
many of them had a chance to do much of anything, the famed English man of letters Samuel
Johnson wryly commented on the rising number of engraved maps of North America: The war
“has incited us to survey and delineate the immense wastes of the western continent by stronger
motives than mere science or curiosity could ever have supplied.”
45
John Campbell the 4
th
Earl
of Loudoun was especially interested—during his brief tenure as the commander-in-chief of
British forces in North America from 1756-7—in assembling cartographic knowledge as a
broader set of military reforms. It seemed this was mostly to aid the planning and execution of
military expeditions. Loudoun, for example, ordered engineers Samuel Holland, John Montresor,
and Charles River to copy and synthesize maps of the Hudson River and Lake Champlain so as
to show this overland route between British and French colonies.
46
We must note a discrepancy between commanding officers and the actual producers of
geographic knowledge in how they valued or thought about accuracy. As a general rule, officers
cared more about aesthetics than accuracy. Engineers understood this and often complied with
this demand for well-drafted maps that dressed up terrain in pleasing ways. As Marshall notes,
many engineers and surveyors “were reluctant to submit work which was not an artistic
rendering.”
47
These engineers and surveyors did not necessarily sacrifice accuracy, as they understood
it, at the altar of aesthetics. Indeed, the few records that directly capture the voices of British
44
Marshall, “The British Military Engineers.”
45
As quoted in Edelson, New Map of Empire, 65.
46
G. D. Scull, ed., The Montresor Journals (New York: Printed for the New-York Historical Society, 1881);
Johnson, Charting the Imperial Will.
47
Marshall, “The British Military Engineers,” quote on 213.
30
surveyors suggest accuracy was a preoccupation, at least in their criticism of maps made by
others. Some cases reveal how closely they had studied particular maps coupled with their
visible disappointment when those maps omitted or misrepresented. In his journals, for example,
John Montresor commented to himself about the accuracy and completeness of existing maps as
the British fleet sailed up the St. Lawrence River to siege Quebec. It bears noting that Montresor
did not seem to intend these journals as public documents for he wrote many impolitic things that
could have badly damaged his career. We can reasonably assume that his criticism was not part
of some rhetorical strategy to bolster his authority at the expense of somebody else—but as close
to an unfiltered, unmediated opinion as a primary source will ever yield.
In some cases, Montressor spoke of many maps: “The Island of St. Pauls in some
Draughts is but just marked, in others not taken notice of, no further than a Dot and the writing
signifying what it is, but ‘tis very erroneous . . .”
48
And in other cases, he singled out specific
maps: “There is a long ledge or Spit that runs S.S.W[.] from Hare Island four miles with some
shrubs growing on it not inserted in Jeffries Chart of the river St Lawrence.”
49
It also becomes
clear reading Montresor’s journal that just because British navigators possessed and studied a
given map, did not mean they then followed its advice: “The winds very light and I could not
find any one vessel Draughted nigh the Shore nothwithstanding the several Draughts or Charts of
the River St Lawrence expressing that circumstance that it must require a leading Gale to pass
through.”
50
Despite the challenges of navigating up the St. Lawrence River, the British fleet arrived
at Quebec’s doorstep in late June 1759. A siege lasting nearly three months ended with a British
victory on the Plains of Abraham where the British were able to draw French troops out of the
48
Scull, ed., The Montresor Journals, 197.
49
Scull, ed., The Montresor Journals, 202,
50
Scull, ed., The Montresor Journals, 205.
31
city. In the winter of 1759-60, Amherst planned what would become a successful three-pronged
expedition Montreal (two armies leaving from Albany, one via Lake Champlain and the other
Lake Ontario, and the third from Quebec). The French generals attempted but failed to retake
Quebec during spring 1760 and then, recognizing the hopelessness of their situation, quickly
surrendered in September when British forces met at Montreal. Crucially, for our story, Amherst
agreed to the French governor’s terms of surrender, which would guarantee rights for the French
populace to remain in Canada with their property and their faith intact under the new British
regime. Amherst divided Canada into three military governments (Montreal, Trois Rivieries, and
Quebec), leaving an excessive seventeen battalions in case they were needed to subdue the
French population. This availability of soldiers coupled with the British metropolitan desire to
visually commemorate the victory and the strength of empire in North America probably helps to
explain the genesis of the project to survey Canada.
51
3.3. Surveying, Drafting, and Gifting The Murray Atlas
Few records survive that directly discuss the technical aspects of surveying and drafting
the Murray Atlas; nothing that gives us much insight into the tools, record keeping, and
symbological practices employed. So while archival investigation reveals no outright falsehoods,
neither does it buttress the Murray Atlas’s visual rhetoric of accuracy. What we can reconstruct
comes from bureaucratic correspondence, mostly between Murray and his superiors: Jeffrey
Amherst, commander-in-chief of the British Army in North America; William Pitt, the Secretary
of State for the Southern Department; and the Board of Trade. These letters hint at three
objectives behind the cartographic project. First, to aid a potential reinvasion of Canada should
the colony be returned to the French during peace negotiations; second, to help govern a
51
Fred Anderson, Crucible of War: The Seven Years’ War and the Fate of Empire in British North America, 1754-
1766 (New York: Alfred A. Knopf, 2000).
32
potentially unruly conquered population; and finally, to help Murray and others in negotiating
the patronage politics of advancement within the British military bureaucracy. I argue that
historians have overestimated the significance of these first two points, failing to see how they
were in many ways subjected to the third patronage rationale.
Two sources provide direct insight into the tools employed during the survey. First in a
letter to Amherst in February 1761, Murray wrote that “We have prepar’d plain Tables and all
requisites” for the survey.
52
The “requisites” here may have referred to an object such as a
surveyor’s compass or geometric quadrant, which mounted on the plain (later “plane”) table
could measure angular distance. Murray’s comment (“We have prepar’d”) suggests that the
surveyors worked with makeshift tools, crafted locally. Second, one of the surveyors Samuel
Holland recalled in 1768 that they had measured distances with chains.
53
The method was
relatively simple and not new in 1760: Measure the distance between two points using a chain;
from both of those points, measure the angle to a third point; finally, use trigonometry to
determine the distances to that third point (this last step is not necessary to do in the field, which
speeds the surveying). In the process of measuring distances and angles, the surveyors also
“drew the visible features of the landscape” using paper mounted on the plain table.
54
These instruments had their limitations. Chains expanded in heat and contracted in cold.
Meanwhile, the tools to measure angular distances probably only subdivided to the degree. These
issues would have detracted from the accuracy of the survey. Yet the simplicity of the tools
“permitted relatively unskilled assistants to accomplish a fairly rapid survey.”
55
The lack of an
identified projection system suggests the surveyors had perhaps not squared their on-the-ground
52
Murray to Amherst (22 Feb 1761), TNA, WO 34/1, 36.
53
Marshall, “The British Military Engineers,” 223-224.
54
Edelson, New Map of Empire, 82.
55
Shields, “Murray Map Cartographically Considered,” 63-64.
33
measurements with established theories concerning projection of three-dimensional objects onto
flat surfaces or more recent debates over the true shape of the Earth.
56
The result, in the
estimation of Shields, was “an improvement in accuracy . . . from earlier maps [of the St.
Lawrence]” but nevertheless “considerable localized error resulting from the inexperience of
surveyors and draftsmen, not to mention the quality of instrumentation.”
57
GIS-aided analysis of
spatial distortion will allow us to sharpen these conclusions but we might add that the objectives
of the survey may have mitigated the need for a high degree of spatial accuracy, as understood
through a Cartesian grid.
Murray spoke to the survey’s first objective—reinvading Canada should it be returned to
France in peace negotiations—in a May 1761 letter to Secretary Pitt, assuring that this “copy of
the survey of Canada” will mean that “happen what will[,] we never again can be at a loss how to
attack and conquer this country in one campaign.” Pitt’s copy, which depicted the “RIVER St.
LAURENCE from MONTREAL TO BERTHEIR,” can be seen in Figure 11. Murray may have
earnestly meant these words or he may have had his private doubts. Whatever the case, by
promising Pitt a copy of the survey at the same time he explained its utility, Murray was framing
himself as a diligent, dedicated, and forward thinking military officer. Months later, Murray
revealed to Amherst that his brother at “court” was attempting to negotiate for Murray a
commission in “an old Regiment.” In this light, we might think of Murray as attempting to
cultivate a patronage relationship with Pitt, using the map as both a gift to the Secretary (marked
by traditional trappings of patronage as evidenced in Figure 12) and a demonstration of his
efficiency.
58
56
Jerry Brotton, A History of the World in 12 Maps (New York: Viking Adult, 2013), 294-336.
57
Shields, “Murray Map Cartographically Considered,” 151.
58
Murray to Amherst (7 Feb 1762), TNA, WO 34/1, quote on 220; Reginald Henry Mahon, Life of General the Hon.
James Murray: A Builder of Canada (London: John Murray, 1921), quote on 288.
34
Figure 11: Samuel Holland and [John] Montresor, “A PLAN of the RIVER St. LAURENCE
from MONTREAL TO BERTHEIR ON THE NORTH and to SOREL ON THE SOUTH SIDE
OF THE RIVER being part OF A SURVEY INTENDED TO QUEBEC” (c.1761), BL, K.Top,
119:29
Figure 12: (Detail from Figure 11) Dedication and Coat of Arms in Pitt’s copy of the Murray
Atlas
Having established Murray’s likely patronage goals, we can now proceed to assess the
map’s utility in light of the stated goal of reinvading Canada. Murray’s emphasis on “how to
attack” suggests he may have been most concerned with documenting terrain and elevation,
conditions that would help officers decide where to place redoubts and soldiers. A cursory visual
35
examination of the atlas reveals a great deal of attention to the terrain—e.g. meadows carefully
delineated from cultivated land; persistent shaded relief throughout.
Variation across different copies of the Murray Atlas however, suggests that we should
be careful to not conflate attention to these kinds of details with accuracy. Figure 13 and Figure
14 show the same area around the Island of Bouchard as depicted in Pitt and Murray’s copies,
respectively. Figure 15 and Figure 16 zoom in even further to show the church of St. Sulpice and
the surrounding terrain. Some differences may have emerged from resurveying after Murray sent
Pitt’s copy—but this explanation is unsatisfying because, of all the copies, Pitt’s comes the
closest in resembling the French land tenure system (visible in Figure 5, Figure 9, as well as
present-day aerial photography, Figure 17) in how the parcels were oriented—although the lots
should have been much thinner.
Figure 13: (Detail from Figure 11) Terrain North of the Island of Bouchard in Pitt’s copy.
36
Figure 14: Same area as in Figure 13, depicted in Murray’s personal copy of the Murray Atlas.
Note the discrepancies in shaded relief and also the additional roads.
Figure 15: (Detail from Figure 11) cultivated land around St. Sulpice in Pitt’s copy. Note the
sharply delineated borders, the continuous depiction of cultivated land (i.e. no meadow
interspersed)
37
Figure 16: Same area as in Figure 15, depicted in Murray’s personal copy. Note the different
geometrical patterns, the depiction of meadow interspersed with cultivated land, and the
undefined borders between landholdings.
Figure 17: Present-day aerial photography of the same area in Figure 13 and Figure 14. Note the
persistence of the older land tenure system (depicted in Figure 5 and Figure 9) into the present,
i.e. thin lots fronting the St. Lawrence River. Source: Esri, DigitalGlobe, GeoEye, i-cubed,
USDA, USGS, AEX, Getmapping, Aerogrid, IGN, IGP, swisstopo, and the GIS User
Community.
We see more continuity between Pitt and Murray’s copies when we focus on other
elements of the terrain such as the forest line (although here too we can spot a few
inconsistencies). At a certain level, then, these discrepancies call into question the survey’s
utility for informing terrain-based tactics. Most likely, in a reinvasion, the survey could have
38
served primarily logistical ends. Features such as roads, dwellings, and windmills—generally
consistent across different copies--would have respectively helped move and feed troops.
59
Route finding figured into Murray’s ambitions as well in that he directed surveyors to
examine lengthy tributaries of the St. Lawrence River and requested of Amherst copies of the
recent survey of Lake Champlain. Indeed he joked in one letter to Amherst that he himself
should “reconoiter the communications from the St. Lawrence to the River St. John and that
River as farr as the fort built by General Monkton . . . Since you [Amherst] can give me no better
employment.” Murray explained, “this will at least amuse me and may perhaps be of some use
hereafter.”
60
Yet the atlas ultimately did not document the overland routes connecting the St.
Lawrence to the Eastern seaboard, suggesting Murray prioritized finishing the map quickly. In
this way, Murray would have probably agreed with Marshall writing two centuries later that the
chief obstacle to British victory in the Seven Years War had been “the nature of geography”
more than deficits in British knowledge of overland routes to Canada.
61
Having a detailed map of
Canada would have certainly helped plan a second invasion but probably Murray knew he was
overstating the case to Pitt in order to inflate the significance of his cartographic
accomplishment.
The map also served administrative purposes—though ones largely misinterpreted by
scholars. In June 1762, responding to orders by the Secretary of State Charles Wyndham, 2
nd
Earl of Egremont (who succeeded Pitt in 1761), Murray and his counterparts in Montreal and
Trois Rivieres sent reports to the Board of Trade on their respective governments. We might note
Murray’s tone of frustration in this report when he lamented that it was “impossible . . . to
59
J. B. Harley, “The Contemporary Mapping of the American Revolutionary War” in Mapping the American
Revolutionary War ed. by Harley, Barbara Bartz Petchenik, and Lawrence W. Towner (Chicago: University of
Chicago Press, 1978), 1-44, see especially 25-35.
60
Murray to Amherst (18 June 1761), TNA, WO 34/1, 92.
61
Marshall “The British Military Engineers,” 199.
39
ascertain exactly, what part of North America, the French stiled Canada” because during the
British conquest “no Chart or Map whatever having fallen into our hands . . . to shew what they
understood by it.”
62
The dearth of records, whether the result of poor French record keeping or
wartime discontinuity, left unanswered geographic questions as basic as where did the colony
begin and end.
It is quite reasonable to assume that Murray wanted the Atlas to answer some of these
basic administrative questions. This would explain why he sent a copy of it with his 1762 report,
which he noted doing in a September 1762 letter to Amherst, (“I have sent a Copy of this Survey
to the Secretary of State to accompany the Report I have been ordered to make of my
government to His Majesty.” This likely referred to the copy seen in Figure 18—fully assembled
but reduced in size—present in the Colonial Office (CO) 700 Series at the (British) National
Archives (TNA), which was once the cartographic storehouse of the Board of Trade. Murray also
noted to Amherst that he had “inserted Plans of the Battle & encampment and attacks.” This
most certainly refers to the next map in the CO 700/Canada series, which can be seen in Figure
19. It depicts the military operations of 1759 in the environs of Quebec.
63
62
Murray, “[Report on the Government of Quebec]” (5 June 1762), TNA, CO 323/15, 183.
63
Murray to Amherst (9 Sept 1761), TNA, WO 34/2, quote on 86.
40
Figure 18: [Murray et al] “[Map in two sheets of the inhabited parts of Canada]” (c.1762) TNA,
CO 700/Canada 18.
Figure 19: [Murray et al], “[Map of Quebec and the environs, with plan of military operations of
1759]” (c.1762), TNA, CO 700/Canada 19.
41
Murray referred extensively to this environs map (Figure 19) in the section of his report
on fortifications, intending the maps to orient the reader in space rather than serve as visual
evidence of the phenomenon he discussed. A wall in poor condition merited comment in the text
of the report (“From K to L is a very bad stockade”
64
—see Figure 20 for this detail from Figure
19) but was drawn on the map in the same manner as all the others. Murray concluded this
section of his report outlining remedies and sheepishly asking for money to implement them (“I
am therefore of the opinion that if His Majesty shall think proper to be at the expense of
strengthening Quebec, the most effectual method will be . . .”
65
). The map coupled with the
report allowed Murray to depict himself as a competent administrator. It might have even
facilitated administration by directing funds where (Murray believed) they needed to go. But the
point here was not collapsing the vast distance across the Atlantic. Murray had not provided an
unvarnished view so that metropolitan officials could make decisions remotely. Instead, he
employed the environs map to buttress and even call attention to his judgment as somebody
operating on the ground.
Figure 20: (Detail in Figure 19) City of Quebec with the letters Murray referenced (“From K to L
is a very bad stockade”) in the example above.
64
Murray, “[Report on the Government of Quebec],” 149.
65
Murray, “[Report on the Government of Quebec],” 151.
42
Curiously, Murray never directly cited or referenced the Survey of Canada in his report
as he did with the environs map of Quebec. It seems, as he whizzed through place names, that he
(again) intended the map to (silently) aid Egremont and the Board of Trade in following along.
Less clear is that he intended the map itself to serve as visual evidence—as some scholars have
assumed. For example, Edelson writes: “These maps made clear how little of the countryside
along the river was occupied and how much flat, open land close to navigation might yet be
cultivated.”
66
In fact, the maps show no such thing—and this is a suspect reading of them. The
extant copies of the survey, including the one held by the Board of Trade, depict dwellings and
road clustered alongside the St. Lawrence River, with only a handful of places void of
settlement, as seen in Figure 21. It does not stand to reason, therefore, that Murray sought to use
the survey to buttress his report’s commentary on the fertility of the landscape, the laziness of its
inhabitants, or the economic schemes he proposed.
Figure 21: (Detail from Figure 18) The copy of the Murray Atlas given to the Secretary of State
included the roads and dwellings (red dots) clustered along the St. Lawrence River that were also
featured in the larger copies.
66
Edelson, New Map of Empire, 83.
43
Another potential piece of evidence concerning the administrative utility of the survey are
notes about each parish (the number of families and men able to bear arms) included on the
version of the Atlas received by the King as well as Murray’s personal copy. For an example of
what these textual notes looked like, Figure 22. Surely this would be useful information for
imperial administrators? Yet it was not included in the copy of the Murray Atlas received by the
Secretary of State and held by the Board of Trade—the people who were drawing up new plans
of government. To summarize, Murray treated the survey as having a limited number of
functions for imperial administration, chief among them orienting overseas officials so that they
could follow Murray’s discussion of specifics. It seems unlikely that he intended the atlas to
directly shape metropolitan ideas about how to structure colonial governance in Canada—only to
bolster his authority on the subject. And for this purpose, the appearance of accuracy mattered
much more than absolute accuracy.
44
Figure 22: An example of notes about parishes (in the King’s copy): “THE PARISH of LONG
POINT . . . This was a new Parish established in 1733 on account of the too great distance of
Point aux Trembles from the Town of Montreal, it extends along the St. Lawrence about a
League and a half upon the Depth of one . . . Families 77 Men able to bear Arms 100.” From
“PLAN of that part of CANADA and the RIVER St. LAWRENCE . . . Survey carried on under
the Direction of the HONOURABLE BRIGADIER GENERAL MURRAY GOVERNOR OF
QUEBEC &c.”” (c. 1762), BL, K.Top 119:24.
Murray and the other people involved in the survey hoped it would help them negotiate
the politics of patronage and reputation so crucial to career advancement in the military. We see
this play out in ways both large and small. One especially telling moment concerns the
production and reception of the aforementioned copy that Murray sent to William Pitt, which
included the dedication to Pitt and his family’s coat of arms. Though Murray sent it to Pitt during
the winter of 1760-1, he mentioned it to Amherst only a year later when it provoked some
embarrassment: “By a letter I had from Lord Egerton the plan must have been shewn to the King
and much approven of but the unnecessary ornament [i.e. the coat of arms] was not liked.” In
explaining himself, Murray simultaneously embraced this gesture and distanced himself from it.
45
In one breath he wrote “I meant it for his [Pitt’s] own use and I thought it a proper compliment
from me” and with another, “I never assumed to myself any merit from the survey.”
67
Murray described a network of people around him who harbored ambitions for his
elevation, ranging from the draftsman who had “officiously put Mr Pitts Arms upon it” to others
who encouraged Murray to “ingratiate” himself “by means of this survey when it is finished.”
And as he usually did, Murray included Amherst in his web of patronage through flattery and
deference. Referencing his friends who wanted him to use the survey to “ingratiate” himself,
Murray assured Amherst that he would not have taken such a “step . . . without your consent,
advice and approbation.” In concluding the saga, Murray pointed to both the vertical (i.e.
subservient) and horizontal (i.e. friendship) relations he shared with Amherst while asking
forgiveness and advice: “I hope Dear Sir you will excuse this Detail & that your being my
commander in cheif [sic] is no reason why I should not consult you as my friend as such I expect
your advice and assistance.”
68
In sum, Murray tiptoed around his ambitions lest he upset
Amherst—but the subtext of his letter was that Murray did indeed have ambitions and believed
the Atlas would help realize them.
Ironically, it seems that Pitt too believed in the power of maps for he did not merely show
but actually gave his copy to the newly ascendant King—hence how it came to reside in the
King’s private map collection (now at the British Library, called the King’s Topographical
Collection). Relations between Pitt and George III were somewhat strained owing to Pitt’s
“difference of opinion with the King’s other ministers as to the immediate necessity of a Spanish
war.” We might wonder if Pitt, before retiring “from business” hoped to remind the King of his
successful management of the British war effort by giving him the visual proof of conquest
67
Murray to Amherst (7 Feb 1762), TNA, WO 34/1, 219.
68
Murray to Amherst (7 Feb 1762), TNA, WO 34/1, 220.
46
emblazoned with his seal. This subtle (and somewhat private) gesture that allowed Pitt, in
Amherst’s words, to go “out in good humour” enough so “to be handsomely rewarded” with a
peerage for his wife and a “£3000 a year pension to himself and his son for both their lives.”
69
Privately, as we have seen, the King grumbled about the “unnecessary ornament” and even had
prepared a second, unsullied copy of the same map.
70
Murray seems not to have quite understood the way Pitt yielded the map or the reasons
for the King’s comments. Instead he took away the lesson the cartophile George III preferred
bare maps. And to this lesson Murray steadfastly adhered. Subsequent copies did not include the
most obvious marks of patronage, sticking rigidly to the Enlightenment aesthetic of keeping that
which was geographically unknown as uncluttered white space. Murray found ways around this
limitation. He was sure to feature his name prominently where he could. The title for the King’s
copy, for example, described the “Survey carried on under the Direction of the HONOURABLE
BRIGADIER GENERAL MURRAY GOVERNOR OF QUEBEC &c.” The King’s copy also
depicted “General Murray’s Encampment” (which can be seen in Figure 23) just outside
Montreal, serving as a reminder of his military service. Perhaps a small detail but we might
wonder if one source of Murray’s anxiety over the copy for Pitt rested in the omission of
Amherst’s encampment (which would have been slightly beyond the map frame, as can be seen
in Figure 24). It all went over the head of Amherst who seems to have not understood the cause
of Murray’s disquiet, namely that the King had taken offense at the map. Instead Amherst
reassured Murray that “Mr. Pittman having Mr. Pitt’s Arms on the Survey can be of no bad
Consequences; Mr. Pitt could not take it any otherwise than a Compliment.”
71
69
Mahon, Life of General the Hon. James Murray, 288.
70
“Plan of the River Saint Lawrence from Montreal to the Parish of Berthier on the North side of the River, and
Sorel on the South” (n.d.), BL, K.Top 119:28.
71
Amherst to Murray (28 March 1762), TNA, WO 34/3, 135.
47
Figure 23: Inset in the King’s copy of the Murray Atlas. Note the reference to the encampments
of Amherst and Murray.
48
Figure 24: Detail from Pitt’s copy (Figure 11). Note that the “white” space of the map begins
immediately after “Gen. Murray’s Encampment” and the absence of “Amherst’s Encampment”
(which following the placement in the King’s copy would rest just to the left of the map frame).
Patronage and reputation emerged again months later in another letter to Amherst,
concerning this same c.1760 copy for William Pitt (Figure 11), but this time concerning two
surveyors John Montresor and Samuel Holland. As Murray explained it, “Montresor says his
Health makes it necessary for him to go to the southern Coloneys.” Murray effectively wrote to
warn Amherst of “a very wrong thing” Montresor had “been detected in doing.” While boxing up
the map to send to Pitt, Montresor “erazd [erased] Holands Name.” Holland suspected as much
for the next day, he “desired that the Box might be broke open and the Plan examined, which
was accordingly done to Montresors disgrace.”
72
When Montresor reached Amherst, the General
(who was “Sorry to find Mr. Montresor . . . Guilty of [this] Ill Judged, wrong piece of Business”)
put him in charge of drafting sketches left by another engineer—perhaps as punishment.
73
The episode, which appears to be true judging from an apparent erasure on Pitt’s copy,
reveals the desperate measures of people like Montresor in claiming full credit to impress a
72
Murray to Amherst (26 Sept 1762), TNA, WO 34/2, 82.
73
Amherst to Murray (28 Oct 1762), WO 34/3, 177.
49
potential patron. By the same token, Holland’s (vindicated) suspicions illustrate that he too
wanted his name to be seen by Pitt, fully aware of the cutthroat competition for advancement.
Ultimately Holland prospered at this game, securing the allegiance of Murray who
enthusiastically endorsed him to whoever would listen including Amherst, Egremont, and the
Board of Trade.
74
This paid off when Holland received a lucrative commission to lead the
northern survey of North America for the Board of Trade.
75
Montresor was left to write a bitter
record about Holland’s salary in this position (“2000 £ per annum Sterling”) to himself in his
diary after he felt his own career had stalled. He titled his list “Mem[orandum]s of British Folly,
not having any good subjects of their own.”
76
Murray may have affected surprise to Amherst about Montresor’s duplicity but he clearly
understood the stakes in placing or omitting names on the final maps. And this wasn’t the only
instance where competition for recognition led to such backhanded tactics. In November 1761
Murray had complained to Amherst about Captain William Spry who had taken charge of “what
he is pleased to call a drawing room” and was contriving “to dislodge Capt. Holland & indeed
every body concern’d in this Affair.”
77
As Shields has argued, much of the rivalry in this
drawing room split along the lines of engineers vs. the battalion officers.
78
It seems possible that
Murray himself fanned these competitive flames as one way to motivate his subordinates. As he
later explained to Amherst:
The Established Engineers and the other officers, who have assisted in taking the
survey and drawing the Plans, have been very diligent, and deserve much praise,
but I found it necessary, to excite emulation and to satisfy the Battalion Officers
74
For example, see Murray to Amherst (9 Sept 1762), TNA, WO 34/1, 86.
75
On Holland’s career, see especially Johnson, Charting the Imperial Will.
76
G. D. Scull, ed., The Montresor Journals (New York: Printed for the New-York Historical Society, 1881), 135.
Scull, ed., Montresor Journals, 135.
77
Murray to Amherst (15 Nov 1761), TNA, WO 34/1, 213-4.
78
Shields, “The Murray Map Cartographically Considered,” 91.
50
that they are not to be rob[b]ed of the honor of their performances, to mark upon
the face of the survey the particular labor of each individual.
79
What does this all add up for us thinking about the value of these and other historical
maps in the context of Geographical Information Science? Primarily, we should take little for
granted in appropriating historical geospatial data from them. To recap this discussion so far,
textual primary sources suggest the overriding rationale for creating and gifting maps in
transatlantic political culture revolved around establishing patronage relationships and building
one’s reputation. This was the first lens through which the more utilitarian objectives of the
mapping were refracted. A second lens, in this particular case, involved the makeshift tools,
techniques, and labor practices that often left something to be desired by contemporary
standards. So we must independently verify the phenomena represented within them as much as
possible before trusting our dataset. This kind of painstaking work will help us understand what
surveyors and draftsmen prioritized, what they fudged, and what they misrepresented. And in
this way, we will begin to understand something about the spatial worldview of these people.
79
Murray to Amherst (9 Sept 1762), TNA, WO 34/2, 86.
51
Chapter 4: Digitally Processing the Murray Atlas
This chapter outlines my methods for digitally processing the Murray Atlas into a GIS as well as
my examination of spatial distortion within the Atlas. Finally, I discuss how I created spatial
interoperability between my historical geodatabase and the demographic records discussed in the
next chapter.
4.1. Creating a Basemap and Vector Data
Digitally processing the Murray Atlas began with the challenge of assembling the
disparate sheets into a single basemap. Slight variations in size, discrepancies in the positions of
geographic features, and warping of the paper over time impeded initial attempts to align
neighboring sheets in Photoshop.
80
Instead, I proceeded by employing the composite map
(Figure 1) as scaffolding for these individual maps—operating under the logic that by working
with the mapmaker’s synthesis of the disparate sheets, I preserved their intent, vision, and
understanding of the geographic whole. In ArcGIS, I georeferenced this composite map with an
affine transformation that maintained its original proportions but nevertheless allowed for easy
comparisons to present-day maps when useful. Next I created vector polygons over the original
grid (refer back to Figure 1), which dictated where each manuscript sheet was located. In most
cases, I used this vector grid to georeference each of the individual manuscript maps, using the
four corners as control points and then enacting a projective transformation. To see what this
process looked like in ArcGIS, see Figure 25. The full results can be seen in chapter one, Figure
2.
80
Paul, P. Bolstad, Gessler, and T. M. Lillesand. “Positional Uncertainty in Manually Digitized Map Data”
International Journal of Geographical Information Systems, no. 4 (1990): 399–412.
52
Figure 25: Assembling the Individual Maps on top of the Composite Map in ArcGIS. See Figure
2 for a view of all the individual sheets assembled together.
Certain sheets provided a special challenge. In some cases the draftsmen, to save space,
included insets—made at the same scale as the main map—of other geographic areas that did not
neatly paste onto the original grid. For an example of what these insets looked like, see Figure
26. In other cases, draftsmen depicted geographic space as spilling over neatlines, again to save
them the task of drafting entirely sheets that would be mostly empty space. For an example of
what this looked like see Figure 27.
53
Figure 26: Example of Insets in the Murray Atlas
Figure 27: Example of Mapped Area Extending beyond Neatline
In the case of insets, I employed Adobe Photoshop to cut and paste the inset to the
appropriate area, which was indicated through text in the inset. Figure 28 shows a screenshot
from Photoshop with the insets being assembled. While not always an ideal solution, it preserved
continuity of intent across different maps and usually aligned with the grid in ways that made
sense as can be seen in Figure 29. Any major distortion that may have resulted in the inset area
from not having additional control points anchored in the grid would later be corrected in the
54
spatial adjustment process. Moreover, in most cases these insets contained spatial data of limited
significance to the primary purposes of this project because very rarely did they depict buildings,
roads, or other elements of the human built landscape. Transgressed neatlines, while more likely
to include more relevant geospatial data, required no additional editing from me. I anchored the
map neatline within the grid, which allowed me to ensure that the overhang conformed to the
logic of the grid as shown in Figure 30.
Figure 28: Assembling Insets in Photoshop
55
Figure 29: Inset Alignment with the Composite Map
Figure 30: Aligning the Neatline with the Grid in Case with Overhang
56
Having created the raster mosaic dataset, I proceeded to conceptualize the geographic
universe being depicted in the Murray Atlas so that I could organize a geodatabase. Here I
created an Entity-Relationship Diagram, which can be seen in Figure 31. My goal was, as much
as possible, preserving the original intent of the cartographer—to see the geographic world in
their terms rather than my own. In this way, creating an Entity-Relationship Diagram was an
empathetic activity, helping me to chart the mapmaker’s worldview.
57
Figure 31: Murray Atlas Entity Relationship Diagram (ERD)
58
Advances in scanning and automation software mean that with relatively clear and legible
paper maps, it is now sometimes possible to recognize certain kinds of geographic features
embedded in the digital scans of those maps and then convert raster into vector data.
81
Unfortunately the Murray Atlas contains too much symbological and typographical variation for
any kind of automated processing. This necessitated painstakingly manually creating vector data
by hand, effectively tracing over the raster backdrop one vertex at a time—a sluggish process
documented in Figure 32, which shows a screenshot in ArcGIS’s data editing mode.
Figure 32: Creating Vector Data in ArcGIS by Tracing Over the Historical Map
There was one upside, however, to this manual processing. Carefully going over inch in
the map facilitated close reading. I encountered details that I otherwise would have certainly
81
Yao-Yi Chiang, Leyk Sefan, and Craig Knoblock, “A Survey of Digital Map Processing Techniques” ACM
Computig Surveys 47:1 (April 2014), 1–44.
59
missed and, ultimately, I really knew what was and wasn’t depicted in the map. Both of these
facts aided my historical interpretation. As one example in the previous chapter (section 0),
knowing the Murray Atlas so well helped me read Murray’s report to the Board of Trade. I knew
that Murray, in discussing the Canadian landscape, was not implicitly referencing any specific
geographic features depicted in the Atlas other than toponyms. This in turn helped me identify
the mistakes of an earlier scholar (Edelson) who believed Murray was using the Atlas to support
an argument laid out in his report.
A key downside to manual vector creation is the potential for human error. To reduce
this, I adopted strategies to ensure constantly rechecking of work and correction of mistakes.
Rather than working on creating all the vector data in one historical map at a time, for example, I
worked vector layer by vector layer across all historical maps. For example, I would process my
“buildings” layer in all maps before moving onto roads or waterways. This also guaranteed that I
employed consistent standards and definitions. If, for example, I had decided to process a certain
visual symbol (a red cross) as a church, this practice ensured that I employed the same standard
throughout. Processing new layers also required me to return to the same geographic areas,
exposing earlier mistakes, omissions, and inconsistencies.
In certain cases, I encoded my uncertainty into the attributes of the vector data.
Sometimes, I noted symbological ambiguities (“is that dot a dwelling or a smudge?”). In other
instances, the uncertainty went deeper. Without the presence of any kind of key or legend,
classifying certain kinds of terrain proved challenging. In these instances, it was often unclear
how much to read into symbological differences. Did circles in what appeared to be marsh
indicate rocks? Or did draftsmen simply render the same phenomenon in slightly different ways?
Rather than answer that question directly, I created subclasses to preserve the symbological
60
distinctions while identifying what was commonplace. In some cases continued scrutiny of the
map helped suggest what certain symbols represented; in other unresolved cases, other scholars
accessing the dataset can choose for themselves if and how to interpret ambiguities.
4.2. Spatial Adjustment
After creating this vector data, I used spatial adjustment to ensure its alignment with our
present-day understanding of the geography. I employed churches as my primary type of human-
created control point, operating under the assumption that they offered the most continuity in the
landscape between past and present. Moreover, judging by the high rank of churches in the
symbological hierarchy, we can conclude that surveyors and draftsmen seem to have paid
consistent attention to churches. In some cases, the churches standing today were built before
1760. In most cases, however, the churches were rebuilt in fairly proximate locations.
As with the process of creating the vector data, I completed this process of locating these
churches manually. On the one hand this reflected a dearth of existing geospatial data—I could
find no dataset of churches providing locations and their building dates. On the other hand, even
if such a dataset existed, it would still need to be carefully checked over to ensure rebuilt
churches did not also relocate. As I had noted uncertainty while creating my vector data, I noted
the evidence (positive and negative) that these historical churches occupied the same location
since 1760. I acquired this evidence through parish websites and Google Maps. In the cases
when a church had not made its architectural history available online, Google’s street view
enabled me to see dates etched in cornerstones, parish cemeteries, and other evidence that
suggested continuity or discontinuity of location. In a handful other instances, I located other
buildings such as fortifications depicted in the Murray Atlas where I could establish a lineage to
61
what stands in the present. Ultimately, I identified eighty-one of these control points. Figure 33
shows their distribution.
Figure 33: Distribution of Control Points for Spatial Adjustment
Yet these alone would not produce a satisfactory transformation owing to their uneven
distribution. This uneven distribution was not especially surprising, having roots in historical
development. Namely, long settled places in 1760 would logically have more buildings than
recently settled places and therefore would be much more likely to have more buildings
surviving into the present. To correct this, I identified an additional fifty-three control points
(again, see Figure 33) rooted in durable qualities of the physical landscape represented in both
the Murray Atlas and present-day aerial/satellite imagery such as islands with distinctive shapes
or places in rivers with distinctive curvatures. Because the objective was to maximize the
alignment between past and present, the adjustment method I employed to transform the vector
data was rubbersheeting.
I had to repeat this process with the original data two or three times to get a satisfactory
result. Here there were two issues. First, poor alignment in a handful of areas around selected
church control points suggested that, perhaps, these particular churches had been relocated
during rebuilding. Removing these control points solved this problem rather quickly. In other
62
cases, spaces between control points did not align well. Finding additional control points in the
durable physical environment solved this problem. The overall result of this work can be seen in
Figure 34 with details of Quebec and Montreal in Figure 35and Figure 36, respectively.
Figure 34: The Digitized Murray Atlas
63
Figure 35: View of Quebec and Environs
64
Figure 36: View of Montreal and Environs
65
4.3. Positional Accuracy and Spatial Distortion
With the historical geodatabase completed, the question now was: What kind of
positional accuracy does the Murray Atlas offer? What is the level of error and how is that error
distributed? To answer these questions, I created a fishnet composed of quadrats over the
unadjusted data and then used the same control points to spatially adjust the fishnet. Next, I used
GIS to measure the area of each quadrat and the length of each side. After factoring in the slight
scale change, I was able to determine the linear distortion for each segment (and areal distortion
for each quadrat). The average linear distortion was -2.06%. 91.2% of the map had less than 15%
linear error while only 2.2% suffered from error greater than 25% (Table 1). The spatial
distribution of that error can be seen in Figure 37 and Figure 38. Said differently, the vast
majority of the map was not very distorted while only a tiny minority was extremely distorted.
Overall, the low degree of spatial distortion outperforms our expectations given the particular
historical context outlined in Chapter 3.
Table 1: Range of Error in Quadrat Lines
Linear Distortion Number Percent
Less than 5% 10065 50.6%
5- 15% 8011 40.2%
15%-25% 1395 7.0%
25%-35% 240 1.2%
Over 35% 197 1.0%
66
Figure 37: Linear Distortion in the Murray Atlas.
Figure 38: Linear Distortion Greater than 10% in the Murray Atlas
More potentially interesting for our purposes: What might the spatial distribution of that
distortion tell us? Visually we can see both sides of the Atlas suffer from large swaths of
distortion while the areas in-between have a few troublesome but confined spots. Some of this
distortion probably emanates in part from the synthesis of the individual map sheets—for
example, some low-grade distortion across downstream portions of the St. Lawrence River,
which is visible in Figure 39. Here we note that the second to last column in the composite map
was slightly wider (2.78 inches) than the others (~2.35 inches)—hence why we would see this
67
area shrink during the spatial adjustment. Figure 40 shows this disparity in column width on the
composite map.
Figure 39: Linear Error (Greater than 10%) in the Eastern Portion of the Murray Atlas
Figure 40: Row from the Rightmost Portion of the Composite Atlas (Figure 1). Note the slightly
longer second-to-last column (compared with the others).
Another important note is that the territory in the government of Quebec was on average,
less distorted (-0.88%) than that in Trois Rivieres (-3.41%) or Montreal (-3.02%) upstream.
Indeed, the distribution of this error, which can be seen in Table 2, also favors Quebec. And from
68
a historical perspective, this all makes sense: The lengthy recon naissance and siege of Quebec
in 1759 probably provided surveyors with preliminary data and greater familiarity with this
terrain.
Table 2: Distribution of Error by Government
Error Montreal Quebec Trois Rivieres
# Quadrat
Segments
% Quadrat
Segments
# Quadrat
Segments
% Quadrat
Segments
# Quadrat
Segments
% Quadrat
Segments
Less than 5% 2664 39.8% 5203 59.2% 1538 45.9%
5%-10% 2197 32.8% 2191 24.9% 1182 35.2%
10%-20% 1368 20.4% 1235 14.1% 458 13.7%
20-30% 324 4.8% 72 0.8% 139 4.1%
Higher than
30%
138 2.1% 82 0.9% 37 1.1%
TOTAL 6691 8783 3354
How did surveyors contribute to the distribution of error? Here we might revisit Shields’
conclusions. He discussed error and surveyors with a circular logic, essentially giving primacy to
the archival evidence. In discussing the Isle d’Orleans where he found high error, he commented
that it leads “one to believe that the survey was not taken in a competent manner.” He continued:
“Interestingly, the surveyor in question is none other than William Spry, who had so irritated
Murray with his alleged ineptitude.”
82
In contrast, Shields downplayed error when committed by
those more favored by the historical record. For example, he wrote: “The Beauport region,
surveyed by Holland, is the area of most serious localized error” which he explains with the
hypothesis that “Holland may have trusted subordinates with the work.”
83
My own results, which
can be seen in Figure 41, tell us that error in both areas was under 10% in contrast with the
82
Shields, “Murray Map Cartographically Considered,” 141-142.
83
Shields, “Murray Map Cartographically Considered,”142.
69
66.4% for Beauport and 35.9% for Île d’Orleans that Shields reported. But what about the
general principle that some surveyors were more prone to significant error than others?
Figure 41: Absence of Linear Distortion Above 10% from Beauport and Île d’Orleans
In our small sample of surveyors, we find a moderate correlation (R
2
=0.4592) between
the average linear distortion and the amount of territory surveyed.
Table 3 lists the average linear error for as well as territory covered by each surveyor as Figure
42 charts this relationship. Figure 43 maps surveyor coverage while shading the places with
more than 10% linear error. Said plainly, surveyors who did more mapping were somewhat more
likely to make mistakes. Other factors such as skill must have also contributed. Samuel Holland,
for example, who had surveyed the most territory and received generally glowing reviews from
superiors had a lower average linear error rate than did three of his counterparts. John
Montresor—the surveyor who attempted to erase Holland’s name and bitterly complained his
diary about the inaccuracy of other mapmakers—only surveyed slightly more than two others yet
had an error rate second only to Fusier who mapped five times more territory. Limitations on
mobility probably also posed problems for surveyors in certain terrain. We might hypothesize
70
that marshes and inland settlement patterns added to the high linear distortion around the
peripheries of Lac-Saint-Pierre, which is shown in Figure 44.
Table 3: Surveyor Coverage and Error (Colors in the table match the key in Figure 43)
Surveyor Area Surveyed (SqKm) Average Linear Error (%)
William Spry 343 3.62
Frederick Haldimand Jr. 374 3.43
John Montresor 481 7.12
Joseph Peach 756 3.78
Philip Pittman 1032 6.72
Lewis Fusier 2703 8.75
Samuel Holland 2780 6.46
.
Figure 42: The Relationship between Surveyor Coverage and Distortion
71
Figure 43: Surveyor Coverage and Error in the Murray Atlas
72
Figure 44: Error around Lake St. Peter with Dwellings in Green
We might investigate one last hypothesis concerning spatial distortion in the Murray
Atlas. Did densely settled areas suffer from less distortion than places with little or no settlement
(i.e. as depicted in the maps)? Visually it would seem so: Areas with highly clustered dwellings
around Quebec and Montreal Cities look to suffer from low linear distortion. This might also
help explain the high linear distortion upstream of Montreal and along the Richelieu River with
fewer dwellings. I tested the proposition by creating points along the perimeters of the quadrats
(as shown in Figure 45) and taking two pieces of data: First the linear distortion for the quadrat
segment and second, to measure clustering, the sum of the inverted distances of dwellings (this
weighs closer dwellings more than further ones) within a 5km radius (a distance that can be
walked at a typical pace in an hour). Surprisingly, I found absolutely no correlation (R
2
=0.0002)
between the clustering of dwellings and distortion. See Figure 46 for a scatterplot showing the
relationship between clustering and distortion. Measuring for density rather than clustering did
not substantively change my results (R
2
=0.00187) as shown in Figure 47. In other words,
73
surveyors did not make more effort to take good measurements because an area was well
settled—this seems as good a clue as any that policing a conquered population was not the
predominant concern of the surveyors.
Figure 45: Points to Sample Linear Error and Clustering
Figure 46: Scatterplot Showing Clustering (Sum of Inverted Distances to Dwellings within 5km)
vs. Linear Error
74
Figure 47: Scatterplot Showing Density (Number of Dwellings within 5km) vs. Linear Error
4.4. Boundaries of Enumeration
Having assessed the spatial accuracy of the Murray Atlas, we might begin to judge its
broader reliability. Did the phenomena depicted within actually exist? Before we can address this
question, we have a few more steps to follow. Specifically, we are testing whether the number
and distribution of dwellings corresponds with other demographic records, namely the censuses
and parish registers recording yearly vital events (births, deaths, and marriages). Before dealing
with those sources, however, we need to ensure their spatial interoperability with our historical
geodatabase. What does that mean, exactly? If the Murray Atlas depicts a dwelling in a particular
location, we need to know the parish—the smallest geographic unit of enumeration for the
census and parish register—to which that dwelling belongs. Otherwise, we could not compare
our different source bases.
The first problem is that the location of administrative boundaries is unclear. The Murray
Atlas does not actually depict them. Meanwhile, the other demographic sources provide very
limited spatial information. When a parish register describes a vital event such as a birth or
75
death, we do not know the specific geographic coordinates at which the event occurred (nor does
it provide an address, for example). When a census describes the population of a parish (the
smallest geographic level of enumeration) we don’t know the precise area where that population
would be bounded in the Murray Atlas. Essentially we can only locate the parish through its
name—so that like creating the vector data, this too is very much a manual process. Even here
we run into some problems: parish churches are usually but not always depicted in the Murray
Atlas; sometimes parishes have multiple names or very long names which are abbreviated in
different ways in different sources; sometimes the parish church does not share the same name as
the parish; and finally, we see parish names given in both French and English.
Fortunately—having anticipated this problem—I folded the work of locating the
coordinates for the parish churches into my research for the spatial adjustment. However
extrapolating parish boundaries from those coordinates proved an ultimately elusive goal. In
some senses, such a task was always a bit of a fool’s errand. From a historical perspective, we
might take a big hint from the absence of depicted administrative boundaries in this, one of the
most obsessive maps drafted before 1800. Here, we should heed the recent argument of Jordan
Branch that the territorially sovereign state and its geographic administration operated much
more in theory than in practice during this period.
84
So if we could create parish boundaries, we
would have to conceptualize them as extremely fuzzy and fluid.
In addition to these theoretical challenges, there were also practical problems. While the
Atlas did not depict administrative boundaries, it textually described the dimensions of many
(but not all) parishes in the margins of the individual map sheets (See Figure 22 in chapter three
for an example of this). These descriptions often noted a length and a width given in leagues.
84
Jordan Branch, The Cartographic State: Maps, Territory and the Origins of Sovereignty (Cambridge: Cambridge
University Press, 2015).
76
The descriptions, however, provided neither anchor points from which to measure these
dimensions nor clarification as to the nature of these distances (e.g. as the crow flies vs. the
contours of the St. Lawrence River shoreline). It is also unclear whether these are English or
French leagues, which would make a difference in their length. Other methods for locating these
parish boundaries yielded little in the way of clarity. I created Thiessen polygons
85
from the
parish points, operating under the logic that people would belong to the most proximate parish
church. I looked at present-day municipal boundaries along with the boundaries depicted in the
surviving portions of the 1709 cadastral survey described earlier (see Figure 9 and Figure 10 in
chapter two). But, as Figure 48 illustrates around the environs of Quebec, ultimately too much
disagreement existed between all of these different sources in too many cases for it to be possible
to create parish boundaries with any kind of confidence.
85
Thiessen polygons partition space around a set of points in such a way as to indicate the closest of those points
from any location.
77
Figure 48: Some Boundary Options. Present-day municipal boundaries (red) and Thiessen
polygons (purple) derived from parish churches (green). Both sets of boundaries often seem at
odds with the human terrain (dwellings in black, cleared land in light gray) as depicted in the
Murray Atlas.
Ultimately then, there was no one good metric of where one parish began and the next
ended. Instead, I aggregated parishes together into larger units (which I call “regions”) as the
way to reconcile the Murray Atlas with parish registries and historical censuses. Here I took
advantage of the fact that some parishes were clearly separated from their neighbors by
discontinuities in settlement—and that in some of these cases, one or more of the sources
described in the paragraph above suggested a boundary existed somewhere within that zone of
discontinuity. See Figure 49 and Figure 50 along with their captions to see how I created region
boundaries at the head of Lake St. Peter and around Montreal Island, respectively.
78
Figure 49: Creating Region Boundaries using Thiessen Polygons. In some cases, I drew the
region boundaries (red) to coincide with Thiessen polygons (purple), such as here where the line
lay between two regions of cleared land (light gray).
Figure 50: Creating Region Boundaries using Natural Barriers. In other cases, I used natural
boundaries such as waterways to establish regions. Here for example, we see two regions (Isle
Jesus and Montreal Island) derived from their status as islands.
79
This is not a perfect method. It relies on the assumption that the map accurately depicts
settlement data. This is not the worst problem because we are testing how well the settlement
data corresponds with other quantitative sources. If the map did a poor job in this regard, this
method of spatial enumeration would create or aggravate inconsistencies between the map and
the other demographic sources—and those inconsistencies would then be visible to us. In other
words if the assumption is bad, then so is the underlying data.
Other problems with this method for creating regions involve a high level subjectivity
that renders it difficult if not impossible to reproduce the same results, i.e. two researchers could
very easily draw different sets of boundaries or group parishes together differently. This made it
essential for me to document how I had reached my conclusions. First, I created a “boundaries”
feature class (line) to document my rationale for each boundary segment that would supplement
the final “regions” feature class (polygon). Figure 51 shows what this documentation process
looked like in ArcGIS with an example of a region boundary within the government of Quebec.
Additionally, I included the disparate data sources informing my decision-making in the final
version of the geodatabase. My last step in the process of making the Murray Atlas spatially
interoperable with the censuses and parish registers involved placing each region into an even
larger unit of enumeration (a “government”) to mirror the administrative division that existed
during the period in which the Murray Atlas was created. The final result of this boundary-
making can be seen in Figure 52.
80
Figure 51: Documenting Sources and Rationales for Boundaries in the Geodatabase.
81
Figure 52: Governments and Regions of the St. Lawrence River Valley, c.1760-1765
82
Chapter 5: The Reliability of the Murray Atlas as a Demographic Source
In this chapter, I assess how completely the Murray Atlas recorded the presence and distribution
of dwellings. In order to do this, I had to draw upon data sources that had independently recorded
similar information. Ideally, this would be a source tallying the number of dwellings but no such
evidence exists (rarely is available historical evidence ideal). Instead, I worked with sources that
could speak to population, specifically censuses and parish registers. Censuses count the number
of something (usually but not necessarily people) at a given moment in time while the parish
registers recorded vital events (in this case births, deaths, and marriages) across time. In this
chapter, I offer some brief context about these two source bases, which informed how I worked
with them. After discussing my methods, I explain my results.
5.1. Historical Context of Early Canadian Demographic Sources
Only in the nineteenth century did the field of statistics begin to emerge as we think of it
today, defined in theory though not always in practice by standardized techniques, methods, and
procedures.
86
We must therefore appreciate the limitations and flaws of earlier demographic
records, especially when specific documentation is thin or non-existent.
In Canada, French governors and religious leaders began informally counting the
population of the colony from the time of settlement, providing overall head counts in occasional
letters and reports. The small size of the population (under 250 according to counts in 1608,
1613, 1620, 1622, 1628, 1629, and 1641) facilitated this process. More formal censuses began in
the 1660s when Canada transitioned from being a corporate colony to a royal one. These could
be quite detailed, providing breakdowns by age, gender, occupation, and geographic location.
Some also provided economic and environmental information concerning subjects such as
86
On the emergence of this discipline, see Hudson, History by Numbers, 26-50.
83
livestock, land cultivated, or crops exported.
87
The complete lists (or rolls) of names survive for
three of these censuses (1666, 1667, 1681) from which the briefer tables were derived. For an
example of what a census roll looks like, see Figure 53, which shows a page from the 1666
census. For an example of what this looked like when compiled into a table, see from the 1685
census. We might guess that the process of data collection (i.e. gathering these names) was the
primary source of error in census taking but one study notes serious discrepancies between these
rolls and their returns. In other words, compiling data introduced further error.
88
87
See Censuses of Canada, 1665 to 1871 (Ottawa: I. B. Taylor, 1876).
88
Hubert Charbonneau and Jacques Légaré, “La Population Du Canada Aux Recensements de 1666 et 1667”
Population 22:6 (1967), 1031–54.
84
Figure 53: Page from Roll of Names for the Census of Canada (1666), ANOM, COL B1F109, 7.
Digitized Version available at LAC (MIKAN no. 2318856).
85
Figure 54: “Recensement général du Canada” (1685), ANOM, FM, COL461, 3.
86
As with mapping, formal census taking in the St. Lawrence River Valley declined in the
eighteenth century. Only one census (1739) was conducted at the parish level. A handful of
others (1719, 1720, 1721, 1734, 1754) provided information for the entire colony. It bears noting
that scholars have not assessed the quality of these censuses taken during the French era.
Moreover, only the tabular results for these censuses survive and not the more comprehensive
rolls so that we have no way of ascertaining or eliminating the compilation error discussed in the
paragraph above.
89
Generally, the British state lagged behind the French one in developing an information
apparatus. That was true of the census too—with the 1086 AD exception of William I’s
Domesday book. In the 1750s intellectuals in the Royal Society were discussing the idea of a
national census but for half a century, popular opposition—rooted in fears over taxation and an
intrusive government—sunk Parliamentary bills to count the population.
90
That did not stop
censuses in the British colonies, however, carried out by governors responding to requests for
information from the Board of Trade. It is more than possible, then, that officials conducting
censuses in Canada had previous experience doing so—or could access people with knowledge
of the process.
91
Immediately after conquest, the British conducted five independent censuses in Canada
within the five-year period of time from 1760 to 1765. The first one came immediately after
military victory during a broader process of securing allegiance from the conquered population.
Here General Amherst instructed various subordinates to go to particular parishes where they
would “not only . . . tender the Oath of Allegiance to the New Subjects, but also to disarm them.”
89
Censuses of Canada, 1665 to 1871.
90
D. V. Glass, Numbering the People: The Eighteenth-Century Population Controversy and the Development of
Census and Vital Statistics in Britain (Farnborough: D. C. Heath, 1973).
91
Robert Wells, Population of the British Colonies in America Before 1776 (Princeton: Princeton University Press,
1975), especially 16.
87
Furthermore they would demand from the “Captain or Commanding Officer of the Militia . . . a
List of the Officers & Men of the militia under his Command & of the number of arms they are
possessed of; Likewise a List of the names of the Inhabitants of said Village or Parish.”
92
The
results of this census were reported in the Murray Atlas, which counted the number of families
and the number of men able to bear arms.
The second round of censuses came in 1761-2 as the Board of Trade was deciding how to
form the new government of the colony. The Board requested reports and censuses from the
three military governors of Quebec, Trois Rivieres, and Montreal, who each sent a census to the
Board in 1762.
93
The rolls for Quebec survive and have been published—though nobody has yet
checked the consistency between them and the returns sent to the Board of Trade.
94
The final
census (1765) was the most comprehensive in terms of the information it sought—but there is
little in the way of archival context explaining how or why it was taken.
95
Parish registers form a second crucial source for reconstructing the demographic history
of Canada. Unlike censuses, which sought to capture the full extent of the population at one
moment in time, the parishes recorded vital events—in Canada these were baptisms, marriages,
and burials—as they happened. These were textual documents that recounted not just the event
but also the particulars surrounding it. A baptismal record, for example, could include details
such as parents; a marriage record might note the person’s place of birth; a burial record would
perhaps name the deceased’s children. The Canadian baptismal and burial records noted not just
92
“[Amherst’s Orders to Capt McIntosh, Capt Steiner, and Lieut Sturling]” (15 Sept 1760), TNA, WO 34/85, 103-
107.
93
James Murray, “[Report on the Government of Quebec]” (5 June 1762), TNA, CO 323, 15, ff. 136-186; Ralph
Burton, “[Report on the State of the Government of Trois Rivieres]” (April 1762), TNA, CO 323/16, 1-14; Thomas
Gage, “[Report on the State of Government of Montreal]” (March 20, 1762), TNA, CO 323/16, 32-51.
94
RAPQ, V:1-143.
95
For the returns of this census, see Censuses of Canada, 1665 to 1871. For the rolls of Trois Rivieres and Montreal,
see RAPQ, XVI:1-121. For the a partial list of names from Quebec, see LAC, RG4-A1, volume 11, pages 4496-
4500.
88
the date of the event but also of the birth and death date (respectively). For a sense of what these
records looked like, see Figure 55, which documents the baptism of one François Plouf in the
parish of Contrecoeur.
Figure 55: François Plouf’s baptism record from the Contrecoeur parish register
(Accessed 28
Nov 2017: https://www.genealogie.umontreal.ca/en/exemple/reconstructions-familiales)
It bears noting that across the early modern period parish registers did not necessarily
record everybody who theoretically fell within their geographic reach. As studious readers above
may have guessed, infants who died before receiving baptism may not have been recorded. Also,
parish registers generally recorded only the life events of people of the church’s faith—so that
(self-acknowledged) Protestants might theoretically not be included in the register of a Catholic
church. Most evidence suggests that Canada was religiously homogenous during the French
era—but it is possible religious dissidents existed but were not recorded as such. The crucial
89
point here is that this is a source base more likely to undercount people rather than over count
them.
96
A second difference between parish registers and the census concerns their location. The
registers were decentralized as a matter of necessity, each one stored at a different church. This
can make it difficult for historians to work with—unless somebody else has expended the time
and energy to gather, compile, and process them in some way. In our (fortunate) case, the
Programme de recherche en démographie historique (PRDH) at the Université de Montréal
began this work compiling parish registers in 1966.
97
They have made some of the results of
their dataset publicly available, namely the tallies of baptisms, burials, and marriages in each
parish for each year when records survive. I did not have access to their propriety database for
this project but it bears noting that their labors resulted in more than the transcription of these
records. They created “individual files” for everybody mentioned in the parish registers, thereby
linking together disparate life events. Figure 56 offers the PRDH’s visualization of an
“individual file” with the different types of information included. It bears noting that while
dozens of books and articles have emerged from this dataset, scholars have not yet applied a
geographic lens to it.
98
96
Hudson, History by Numbers, 14
97
For a brief history of the PRDH, see “About the PRDH” (2017). Accessed 30 Nov 2017:
https://www.genealogie.umontreal.ca/en/le-prdh. For an early statement of the scope of their project, see Jacques
Légaré, “A Population Register for Canada under the French Regime: Context, Scope, Content and Applications”
Canadian Studies in Population 15:1 (1988): 1–16.
98
For the full bibliography as compiled by the PRDH, see “Bibliography of the PRDH” (2017). Accessed 30 Nov
2017: https://www.genealogie.umontreal.ca/en/bibliographie
90
Figure 56: PRDH’s “Individual File.” This ties together entries in the parish registers through the
same individual (in this case François Plouf mentioned in Figure 55). Accessed 28 Nov 2017:
https://www.genealogie.umontreal.ca/en/exemple/reconstructions-familiales
5.2. Potential Pitfalls in Working with the Census and Parish Registers
In examining these demographic source bases (and cross-referencing them with the
Murray Atlas), we run into five potential problems that merit discussion. These problems lessen
our ability to precisely judge the quality of the Murray Atlas as a demographic record but they do
not necessarily render it an impossible task. In other words, we will not be able to identify the
exact error rate in one source base as we could do with spatial distortion in the previous chapter.
But we will get a sense of whether the error rate is unacceptably high.
The first problem is that—as with the Murray Atlas—we do not necessarily know that
censuses and parish registers did a good job acquiring the information they sought. The answer
lies in cross-referencing the different quantitative sources. If the data collection methods of at
least two of these various source bases were good enough to adequately capture the underlying
data, we should find they reinforce each other in some key way. It bears noting that cross-
91
referencing different quantitative sources in this way does not always yield ready answers. If
census, parish registry, and map all disagree, that tells us that at least two of these sources are
faulty in some way but it is still possible that one is good. This method will not tell us which
source might be good though we can still resort to logic and interrogate the internal coherence of
the three separate source bases. For example, because we have access to five censuses taken
within a five-year period of time, we can compare the censuses with each other. Finally, it bears
noting that there are some labor-intensive methods by which one could evaluate error rate in the
census and the parish registers. I outline this future work in my conclusion.
Second, as we attempt to reconcile the map with these sources, the interchangeability of
people, families, and dwellings is unclear; we neither know the average ratios of these different
measures in 1760 nor the degree to which those averages varied across space. This problem
requires us to think about a range of acceptable outcomes rather than a single answer that would
vindicate the two sources being cross-referenced. For example, we might be content, if looking
at the Murray Atlas and the censuses, we find that the average ratio of people per dwellings
ranged from five to ten across our different regions. If, however, we find that one region has
twenty people per dwelling and a neighboring region has half a person per dwelling, then we
might suspect one or both sources to be faulty. In other words, we are looking for disqualifying
or suspect results.
Our third and fourth problems are somewhat related: Missing records and inconsistencies
in those that survive. We will begin discussing these problems as they pertain to the census. In
the above section (0) I noted how information gathered under the French Regime (i.e. rolls of
names) did not always make it into the census returns. This was probably an issue with the five
92
censuses in this study. Additionally, the 1765 census did not include information for Montreal or
Quebec Cities.
More problematic than missing information was inconsistency. These five censuses
sought different information, defined categories differently, and did not always operate at the
same geographic level. They usually broke population numbers down by age and gender—but
usually in erratic ways. For example, one census might break the female population down into
girls and women but with no clear age boundary distinguishing the former from the latter.
Another might include a group of people such as strangers or slaves whose gender is not noted.
Some censuses did record specific age boundaries but then the age boundary in one might be
totally different from another. Figure 57, which shows a page from the 1762 Montreal census,
offers one striking case (within the same census!) in which the age boundary distinguishing boys
from men (age 16) was different than the one for girls from women (age 12). For a sense of the
inconsistency between censuses, compare this with Figure 58, which comes from the 1762
census of the government of Quebec.
93
Figure 57: “A General Return for the Government of Montreal” in Thomas Gage, “[State of
Government of Montreal]” (March 20, 1762), TNA, CO 323/16, ff.32-51. Note that this census
only provides population data for the entire government of Montreal. Also note the inconsistency
of categories: It distinguishes boys from men at age 16 but girls from women at age 12.
94
Figure 58: “Number of Souls in the Government of Quebec” in “[Report on the Government of
Quebec]” (5 June 1762), TNA, CO 323, 15, ff. 136-186. Unlike the (above) Montreal census, the
Quebec census provides a population breakdown by parish. Also note some of the internal
discrepancies in the categories here. For example, we have three clearly defined age categories
for males (“Males from 16 to 69” “Males above 60” and “Males under 16”) in contrast with just
two semi-defined ones for females (“Women” and “girls”).
95
These inconsistencies in categorization and missing data limit our ability to fully compare
these three different moments (1760, 1762, and 1765). Only the 1760 and 1765 censuses count
the number of families. Because of the missing data in the 1765 census, it is not possible to draw
comparisons at the colony-wide level (we run into the same exact problem comparing the raw
number of people between the three 1762 censuses and the 1765 one). The governors employed
different metrics in 1762—so for example, we cannot compare directly categories like age
between them and the 1765 census (where they drew the line between child and adult varied).
Also in 1762, the Montreal governor didn’t bother to break down his data by parish so we only
have numbers at the government level. For us, this jumble of early modern record keeping limits
what we can compare. Usually it means that—until we have a better sense of the census
quality—we must work with raw population totals and be flexible about which pieces of the
censuses we compare.
In the case of the Canadian parish registries, the largest frustration grows out of missing
records (i.e. whatever inconsistencies exist in the original textual source do not bear upon the
basic fact of the particular event having occurred). Of the 120 Canadian parishes active before
1760, only 29 have complete records from their date they were established. While the remaining
parishes may have lost an odd year of records here or there, what we see most commonly are
missing records in the earliest years of a parish, i.e. immediately after it was first established.
Therefore the percentage of surviving data usually declines in periods during which large
numbers of new parishes were established. Not surprisingly this effect diminishes as the total
number of parishes increases.
We see the most erratic movement in the percentage of surviving parish data in the
seventeenth century. This is shown in Figure 59, which plots the raw numbers of operational
96
parishes and surviving records from the colony’s founding to 1765. Figure 60 shows the
percentage of records that survive across the same time period. The survival rate begins at 0%
but jumps erratically back and forth to 100% during the years when only one parish (Notre-
Dame-de-Quebec) existed. With a few more parishes established, the rate rises to about 80% in
the 1640s but then hovers just above 50% throughout the rest of the century as state sponsored
immigration contributed to the steady opening of new parishes. Then the number slowly climbs
throughout the eighteenth century, reaching to 90% and higher in the last three decades of the
study period. Overall, this missing data amounts to 20% of all records from 1616-1765,
including from parishes outside the territory depicted in the Murray Atlas (the overall survival
rate is slightly lower for just these parish registers at 78%). For a regional breakdown of parish
register survival, see Figure 61. Our silver lining here is that 98% of parish records survive for
the year 1760. In other words, we have a relatively complete record for the year that matters
most to us. Moreover, the parish record survival stays above 95% through 1765 so that we have
relatively complete data for evaluating the 1762 and 1765 censuses.
Figure 59: Number of Parishes vs. Surviving Parish Records, 1615-1765
97
Figure 60: Percentage of Surviving Parish Records, 1615-1765.
Figure 61: Survival of Parish Registers by Region
The final problem concerns the alignment of map, census, and parish register. In chapter
four, I discussed the spatial dimension of this problem vis-à-vis the boundaries of enumeration.
In an ideal world, we would also be able to create a perfect temporal alignment—meaning we
could translate the vital events in the parish records into a series of temporal snapshots showing
the distribution of population at discreet moments. Yet this problem does not prevent us from
using the parish registers: The strong survival rate of parish records in the years concerning our
study means that we can use the raw number of births as our demographic indicator with which
to check the accuracy of the censuses and the Murray Atlas.
98
Having explained these two specific problems with the parish registers (missing records,
temporal misalignment), I should now add that I have created models to fill in the values of these
missing records—but have saved my discussion of them for the appendices. The first model
estimates the values of births and natural growth (i.e. births minus deaths) for the missing years
in the parish registers (Appendix A). The second model estimates yearly population values for
the entire colony (Appendix B). As I discuss in those appendices, I have some doubts about my
results although I think they deepen how we think about the registers that survive. I discuss the
results of these models a handful of times within the body of this thesis—but explicitly identify
them as such. Otherwise, I stick with the values derived from the surviving registers alone.
5.3. Assessing the Quality of the Censuses
As noted before, we can compare the censuses not just with other demographic sources
but also with each other. However, inconsistencies and some missing data mean that we can only
selectively compare portions of them. We begin by comparing the growing or shrinking number
of families in the 1760 and 1765 censuses across the different regions. Figure 62 maps out this
change and a more detailed table can be found in Appendix C. Our first observation here must be
the remarkable degree of variation from one region to another. For a relatively short period of
time, we see quite large changes. Of the 22 regions, fourteen (nearly two thirds) apparently grew
or shrunk by more than 10%. Table 4 shows the distribution of this change in the number of
families according to the census.
99
Figure 62: Demographic Change According to the 1760, 1762, and 1765 Censuses
Table 4: Distribution of Recorded Changes in Number of Families between
1760 and 1765 Censuses.
% Difference in
# of Families
0-5% 5-10% 10-15% 15-20% 20-30% 30-40% 50-60% 80-90%
Number of
Regions
2 6 3 4 4 1 1 1
We might attribute some of this to internal migration (i.e. families moving from shrinking
regions to growing ones), natural population growth, the potential for statistical misinterpretation
(e.g. an influx of families to a relatively unpopulated region creating having an outsize effect),
100
and the possibility that young people had been putting off marriage during wartime. Looking at
the overall increase in the number of families (around 5%) in this period, however, suggests this
last explanation was not an especially serious factor. The transition from wartime into peacetime
does potentially explain why there might have been a larger degree of internal migration than
normal. Here we run into the Modifiable Areal Unit Problem (MAUP)—probably much of any
hypothetical internal migration occurred within regional boundaries as between them. But why
would we see so much internal migration spread throughout the colony when war did not affect
every region in the colony? Sure, for the sieged Quebec region we might expect to see people
going to the city for protection from British soldiers roaming the countryside and then returning
afterwards. This doesn’t explain the variation in the regions under the governments of Montreal
and Trois Rivieres. Did French colonists really flee so far afield? It seems unlikely.
We readily spot some inconsistencies as we zoom in to particular regions. For example,
we might look to the region of Quebec East where the number of families may have, according
to this census data, nearly doubled in five years from 155 to 291. Yet in comparing the
population difference from 1762 and 1765 censuses for the same region, we only note a shift
from 1451 to 1590 people (a 9.6% increase). Perhaps the remainder of the increase in the number
of families occurred from 1760 to 1762? Whatever explanations we can concoct, the 88%
increase in the number of families still seems beyond likely.
If we turn to the parish registers, here, we do find that the number of births jumps
somewhat dramatically from 68 (1760) to 87 (1761) and eventually reaches 96 (1765)—an
41.2% increase in a five-year period or half the increase we might expect based on the increase
in families. Table 1 provides the full range of data for this decade while Figure 63 plots it on a
graph. We might exercise additional caution, however, in assuming that increases or decreases in
101
births tells us much about population in these years. Here we might note that in the Quebec East
region the raw number of births had been hovering in the low 80s and upper 70s throughout most
of the 1750s so that it had dipped significantly in 1760. In fact this trend mirrors a quite
uncharacteristic drop in births occurring across the government of Quebec in 1758-1760 (which
we can almost certainly attribute to the war and the siege in particular) and then increase after
1761. So the sudden rise of births probably speaks less to a burst of growth but a return to the
status quo and then some modest growth. Most likely, the 87% increase in families resulted from
an error in data acquisition for one or both censuses.
Table 5: Raw number of births in Quebec, 1755-1765, according to surviving Parish Registers.
Region
Year
1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765
Quebec East 79 61 81 73 64 68 87 85 89 89 96
Government of
Quebec Overall
1770 1626 1824 1615 1411 1311 1517 1701 1794 1912 2022
Figure 63: Raw Number Births in the Government of Quebec, 1750-1765, According to the
Parish Registers. Note the significant dip in 1759 for all regions and the continuation of this dip
for Quebec City and Suburbs in 1760. This was almost certainly the result of the British siege
and conquest in 1759.
102
Does our impression of the census change as we look directly at population (rather than
families) by comparing the 1762 and 1765 censuses? Not really. The distribution of regional
change can be seen in Figure 62 above and Table 6 below (for the full table, see Appendix D).
First, we must qualify our source. We cannot include Montreal in our analysis at all because the
1762 census does not give us a parish breakdown of population while the 1765 does not have
data for Montreal City. In other words, we can neither compare individual regions nor the entire
government. In looking to Trois Rivieres and Quebec, the variation is less extreme than when we
measured the family data (which makes some sense because the time frame is briefer). Eight out
of the fourteen regions (57%) grew or shrunk by less than 10% and there is only one (7%) outlier
beyond 30%. But some of this variation is nevertheless perplexing. One region, for example,
apparently grew by about 80% in a three-year period.
Table 6: Distribution of Recorded Changes in Population between 1762 and 1765 Censuses
Percent Difference in Population 0-5% 5-10% 10-15% 15-20% 20-30% 80-90%
Number of Regions 5 3 2 1 2 1
In fact of the thirteen regions for which population data is available in both the 1762 and
1765 censuses, we might note that fully nine (69%) of them tell us the exact opposite as the
families data about whether the region is growing or shrinking in this five year period of time. In
some cases, the difference is not especially large such as in Trois Rivieres North—the number of
families apparently declined by 1% while the population increased by 5%. In other cases, we see
baffling differences: Trois Riveres West (30% vs. -30%), Trois Rivieres South (-7% vs. 80%),
Quebec West (-2% vs. 57%), Quebec City and Environs (29% vs. -19%; note that parish data for
Quebec City is missing for 1765 so was not factored into the 1762 regional data either)—to note
four of the most serious cases. In other cases we find a dramatic difference though not a dispute
103
of the direction of growth or decline. Apparently families in Quebec West grew at a rate of 88%
over five years while population grew only at 10% in three years. The numbers simply do not
add up. The censuses do not support each other or allow us to construct coherent narratives about
demographic change.
That the censuses do not agree with each other does not conclusively prove they are bad.
For this, we might test them according to their Crude Birth Rate (CBR). CBR is a measure for
the number of births per thousand people (calculated by dividing the number of births by the
population and multiplying by 1000). For each region and for the colony as a whole, we will
couple the census data with the number of births according to the surviving parish registers. We
will then ask if these CBRs look plausible or unrealistic. Here we begin by asking: what is a
reasonable ceiling for a CBR?
In a dataset compiled by the World Bank comparing national CBRs from 1960 to 2015,
more than 50% of all countries had a CBR above 40 from 1960 to 1967 (it subsequently declined
to below 50%). Throughout the 1960s, more than 5% of countries (almost entirely in Africa) had
CBRs above 50. The highest CBR in any country between 1960 and 2015 was 58.228—
specifically in Niger in 1962. In the most recent year available in this dataset (2015), Niger still
held the world’s highest CBR at 48.438 while only 3.2% of all countries had CBRs above 40.
The point I am making is that over the past six decades the metrics surrounding international
CBRs have uniformly followed a downward trajectory—owing largely to the greater availability
of effective contraception, rising divorce rates, and the fact that in economically “developed”
countries children carry more financial costs than benefits. If this dataset extended further into
the past, it is theoretically possible that we might find a CBR greater than 58. But other datasets
104
with non-comprehensive pre-1960 records such as Max Roser’s “Our World in Data” do not
document CBRs that are any higher.
99
In this light, a CBR of 58.54 (1762 Census) seems at the limit of what is possible—
though not necessarily likely. A CBR of 78.60 (1765 Census
100
) seems absolutely implausible—
both as an unrealistically high number and for the magnitude of the jump from three years
earlier. We only have two real explanations here: The censuses were undercounting people or the
parish registers were over counting births. Let’s think through the question. Logically, it is
usually much easier to undercount than over count—and we can follow a well-documented paper
trail for the parish registers. The only ways that our parish data would be over counting births is
if priests (or the people compiling the PRDH database centuries later) were fabricating baptisms.
But what would motivate such a widespread deception? Also, given that we are only working
with surviving parish records, we should expect that the number of births (and therefore, the
CBR) should be slightly higher for both years if we incorporated data for the missing parishes.
Therefore, we must conclude that the censuses, particular the 1765 census, were vastly
undercounting people.
The 1760 census counted families rather than people, rendering it slightly more difficult
to test this census using the measure of CBR. But what happens if, as a hypothetical exercise, we
test what the CBR would be if we use different average family sizes? Table 7 shows the
abbreviated results while Appendix E offers a breakdown by region. We find that the average
family size would need to be somewhere around seven to provide a barely possible CBR (60.27).
99
Dirk J. va de Kaa, “Europe’s Second Demographic Transition” Population Bulletin 42:1 (1987), 1-59; Rowland,
Demographic Methods and Concepts, 220-261; The World Bank, “Birth rate, crude (per 1,000 people)” (2017),
Published online at www.worldbank.org. Retrieved from: https://data.worldbank.org/indicator/SP.DYN.CBRT.IN
[Online Resource]; Max Roser, “Fertility” (2017), Published online at OurWorldInData.org. Retrieved from:
https://ourworldindata.org/fertility/ [Online Resource]
100
Because the 1765 census does not record Quebec or Montreal cities, this calculation of CBR excludes the births
recorded (681) from parishes.
105
Only when family size reaches an average of eight does our CBR (52.74) begin to look
somewhat likely if still a little high.
Table 7: Summary of Hypothetical Colony-Wide CBRs for 1760 Census based on Average
Family Size
101
# of Births # of Families
CBR based on Average Family Size of:
4 5 6 7 8
4659 11043 105.47 84.38 70.32 60.27 52.74
Is an average family size of eight plausible in Canada? The 1765 Canadian census
(55,100 people in 10,660 families) would suggest an average family size of about 5.17—though
we have already noted problems with that census. We can also look at the first federal census of
the United States (1790), which provides an average family size for some counties in some states
(Connecticut, Maryland, Massachusetts, New Hampshire, New York, Pennsylvania, Rhode
Island, South Carolina, and Vermont), depending upon whether the original rolls survive or are
missing. Across these counties we find the lowest average to be 4.3 and the highest to be 24.3 in
Dorchester County, MD. However, counties in southern states had much larger numbers, owing
to the aggregation of slaves into the (white) families being counted. The highest rate in the North
was in Kings County (8.3 people/family), which encompassed New York City, and was also
inflated by the incorporation of slaves into white households.
102
If we look ahead to Appendix B,
we find a population estimate from 93,929 to 97,571 by this date. This would suggest a large
average family size of 8.51-8.84. Either the 1760 census takers had a somewhat looser definition
of family than other censuses or they undercounted the number of families. Historical context—a
101
For breakdown by region, see the full table in Appendix E
102
Steven Manson, Jonathan Schroeder, David Van Riper, and Steven Ruggles. IPUMS National Historical
Geographic Information System: Version 12.0 [Database] (Minneapolis: University of Minnesota. 2017).
106
military quickly administering the census while taking oaths of loyalty and securing firearms to a
recently conquered population—favors the latter interpretation.
All in all, we should be somewhat wary of trusting census data. Future work might
involve checking the surviving census rolls against their returns to see what error, if any, was
introduced in originally compiling this demographic data. Scholars may also wish to cross
reference the people in the census rolls with those in the parish registers. This would more
precisely illuminate the limitations of either source base and how much one or both were under
or over counting. Another question to ask: Do the PRDH’s family reconstructions match the
household structure we see in the portions of the censuses for which the rolls survive? This
would be an easy enough question to answer through limited sampling.
5.4. The Parish Data and Murray Atlas as Demographic Sources that Reinforce Each
Other
GIS lets us easily count the number of dwellings in each region as well as the total
number (12,911) across the whole map. Here we must note that the mapmakers did not depict
any dwellings within the walls of Montreal City so this number is a little low. As with the
question of “average” family size in the previous section, we must ask what was the average
number of people per house? Demographers of early North America often use a multiplier of
seven.
103
Indeed, if we look ahead to Appendix B for our overall population numbers (93,929
from just surviving parish registers or 97,571 with estimates for missing registers), we find an
average between 7.28 and 7.56 people per dwelling. This ratio is a little artificially high given
Montreal’s missing dwellings and because these population numbers stretch into unmapped areas
(with more dwellings, presumably). In this light, seven seems a reasonable ratio. Would the
103
Evarts Greene and Virginia Harrington, American Population Before the Federal Census of 1790 (Baltimore:
Genealogical Publishing Co., Inc., 1993), xxiii.
107
distribution of population derived from multiplying the houses in each region by seven produce a
reasonable range of CBRs?
Before answering this question, we must discuss two problematic regions. First, we run
into problems with Montreal Island because the undercounting of houses in Montreal City raises
the ratio of people per house. Montreal West is also problematic though for different reasons. It
contains only one, recently established parish—in other words, a small and unusual sample. It is
likely that, at that point in time, many people here were linked to parishes in other regions,
especially Montreal Island, and therefore registering their births there.
Here if we aggregate Montreal Island with Montreal West, then add a hypothetical 570
dwellings (the number within the walls of Quebec City) to fill in the cartographic omission, we
find a CBR of 57.5—high but certainly not implausible. Another region, Quebec East, has a
slightly higher CBR (60.6). In both cases it is easy enough to explain these numbers. Perhaps
these regions had a slightly higher average people/dwelling ratio (which would lower the CBR).
Or perhaps a ceiling of 58 is high for a country but not necessarily for a smaller geographic unit
of enumeration.
Otherwise, all of the other regions lie within a reasonable range of CBRs as discussed in
the above section (0) on the census. Moreover, the map of CBR rates, shown in Figure 64,
appears to show signs of clustering. We see higher values in the government of Montreal, the
lowest numbers in the government of Trois Rivieres, and a mixture in the government of Quebec
but with generally well-matched neighbors. Unfortunately, it is difficult to test this hypothesis in
a statistically meaningful way given the linear geometry of the map and the inadequate number
of regions.
108
Figure 64: CBR According to A) Population Projections from Dwellings in the Murray Atlas and B) 1760 Parish Registers
109
What happens when we examine migration using these two sources together? Again, we
find that a cohesive narrative emerges. We can calculate net migration between different regions
by comparing natural growth from 1616-1760 with our population numbers derived from
dwellings in the Murray Atlas. If, for example, we look at a region like Quebec City and
Environs where this cartographic method would suggest a population of 13,526 inhabitants vs.
the 9,809 people we would expect to find from natural growth, we can calculate the net
migration to the region (from beyond the colony or other regions) is 3,717 people.
Here we see a narrative about migration that makes intuitive sense. Cursory visual
analysis, offered in Figure 65, shows that the eastern-most regions experienced outmigration
while others absorbed migrants. When we incorporate the numbers from the unmapped parishes
in the Murray Atlas, which largely lay further downstream (easterly) on the St. Lawrence River,
we find a similar exodus (-61.8%).
104
The directional flow of migration is only one part of this story. Time appears to be
another. Figure 66 plots out relationship between the percentage of pre-1700 parishes in a given
region and the net migration to or from that region by 1760. Regions with a larger percent of
parishes established after 1700 were moderately more likely (R
2
=0.4318) to receive a greater
share of migrants than those with more pre-1700 parishes. This is what we would logically
expect to find—i.e. at a certain point older parishes stop expanding geographically and must
either become denser or lose whatever surplus population they might be generating.
104
This data is a little messy. Most but not all of these parishes lie to the east of the mapped territory but we can only
estimate population for them as a whole. We are not deriving this estimate from the houses in these places but rather
by subtracting our population estimate for the mapped areas (95715) from our overall population (97577) estimate.
We don’t know precisely how this unmapped population was actually distributed. We also know that slight changes
to our ratio (people/dwellings) to determine population within the map will significantly change our estimate for this
unmapped population much more than for other regions. For example, changing the ratio from 7.1 to 7.0
people/dwelling, would yield a net migration rate of -34.4% instead of -61.8%.
110
Figure 65: Migration as a Percent of Natural Growth
111
Figure 66: Relationship between Time (Percentage of parishes founded before 1700) and Net Migration (Divided by the Region’s
Natural Growth)
112
This preliminary inquiry suggests that the parish registers and the Murray Atlas support
each other’s demographic accuracy. We have found no obvious contradictions as we did
analyzing the census. Moreover, in conversation these sources suggest a sensible narrative of
internal migration within the colony. Deeper analysis of the people in the parish registries and
census rolls would probably help us solidify this conclusion by elucidating how migration
worked within the colony—which regions produced or absorbed surplus population and when—
and providing us a snapshot of population distribution in 1760 that we could use to compare with
the map. Such future research might tell us even more about how we can use historical maps as
demographic sources.
113
Chapter 6: Conclusions
I began by suggesting that the spatial humanities would benefit from a closer integration of
digital and archival methods, which I have modeled throughout this thesis. Thinking about
accuracy in an expansive undertaking such as the Murray Atlas benefits from pairing historical
context with computational tools. In reviewing the existing literature on the Murray Atlas, I
argued that earlier scholars have misinterpreted the archival sources. In reviewing that archival
evidence myself, I have told stories about maps shaping human geography, the utility of maps
within patronage networks, and the conflicts involving and limitations placed upon mapmakers
in conducting the 1760 survey of the St. Lawrence River Valley. But in many ways, these are by
themselves not complete stories. Digital approaches helped us fill in the archival silences.
Close visual analysis, encouraged by digital processing, helped me to identify
inconsistencies across different copies mainly to do with terrain. This in turn helped me to think
about what geographic features—roads, the forest line, dwellings, churches, mills—surveyors
and draftsmen depicted consistently and, perhaps we can say, valued most. Spatial analysis
conducted in chapter four revealed a low rate of linear error and suggested that temporal
limitations shaped error in two key ways. First, covering a greater amount of territory in roughly
the same time allotted to other surveyors increased the likelihood of error. Second, having
greater familiarity and more time to survey the territory within the government of Quebec
improved the spatial accuracy of the Murray Atlas there as compared with Trois Rivieres and
Montreal. Quantitative analysis of demographic sources in chapter five revealed fatal problems
with the British censuses taken between 1760 and 1765. Yet it also suggested that the parish
registers and the Murray Atlas reinforce each other as records of population. As if recording the
114
results of the 1760 census wasn’t enough of a clue, we can deduce that Amherst, Murray, and the
surveyors working under them were concerned with recording the distribution of people.
The Murray Atlas was by no means perfect—what map is? But—given the personality
disputes, the limitations of time and tools, the extent of the territory surveyed, and the primacy of
patronage goals in guiding the map’s creation and execution—it is surprisingly accurate in two
critical measures. Does this knowledge cause us to revisit our conclusions from chapter three?
Certainly it does. It appears that those who surveyed the landscape did so well and to the best of
their ability. We know from Montresor’s attempt to erase Holland’s name from one copy that
these surveyors—mostly voiceless in the archival record—cared about publicizing their work to
potential patrons. But it seems they also cared about the accurate execution of their fieldwork.
Perhaps they wanted to impress their supervisor Murray who himself knew the terrain as one of
the architects of the invasion of 1759-60 and was perhaps in an unusually strong position to
judge their work.
Combining digital and archival methods provides more insight into how and why the map
was produced than would either approach alone. The resultant geodatabase also provides us with
insight into the landscape it depicted. What new things might this geodatabase potentially tell us
about the historical geography of Canada? I do not mean to constrain the direction of future
research, of course, but I would simply offer a few preliminary thoughts. First, this historical
spatial data help us confirm what we think we already know. Looking at individual map sheets,
for example, it looks like most buildings lie close to larger waterways—the St. Lawrence River
or its major tributaries. Indeed, we can now say that of the 12,911 dwellings in the Murray Atlas,
10438 (or 80.8%) are within 1km—roughly a 10 to 15 minute walk—of a major waterway. In
other words, people valued proximity to the riverfront. This is undoubtedly a story of rivers
115
offering mobility, transportation, irrigation, soil fertility, and a visible landmark by which people
could orient themselves.
The historical geodatabase not only confirms our suspicions but also deepens the kinds of
questions we can ask. Where, for example, are these outlier dwellings, situated more than 1km
away from what is depicted as a major waterway? In looking at Figure 67, we find most of them
in six primary locations, with a range of different explanation why. The environs of Quebec (#4
in Figure 67) boast the most dwellings (909), reflecting the much earlier efforts of imperial
officials from roughly 1660 to 1690 to create a densely populated area around the capital. Across
the St. Lawrence (#4) and much further downstream (#5), we see another two clusters. These
dwellings happen to hug what the mapmakers depicted as slightly smaller rivers. The distance
from major waterways in other areas such as around Lac St. Pierre (#2) or near Trois Rivieres
(#3) suggest that for these French people, environmental barriers (swampy or marshy land) to
establishing such close proximity to a major waterway did not outweigh the value of having a
continuity of settlement. Perhaps being able to geographically web the familial relationships
established in the PRDH database will help us explain why.
116
Figure 67: Dwellings within and beyond 1 Kilometer of a Major Waterway
Finally, on the shore opposite Montreal City (#1) we find a second line of dwellings
forming behind those lining the shore of the St. Lawrence. Why here, particularly? A kernel
density analysis, visible in Figure 68 and Figure 69, helps elucidate the reasons. First, we might
expect Montreal City as a dense urban area to lie at the center of this hotspot but it actually lies at
the edge of it. The explanation for why rests largely in the environment; the width of St.
Lawrence River narrows here slightly compared with areas further downstream while the narrow
wedges of two larger islands converge here. We know from Figure 70 that for the decades
preceding 1760 this region (Montreal South Shore) yielded a high and consistent natural
population growth. Figure 65 suggests the broader region in general absorbed slightly more
117
people than it exported but this number encompasses the entire history of the colony and
includes immigrants to Canada as well as within it. In other words, it seems that this area was
producing “surplus” population. Had that surplus population wanted, it could have relocated just
ten kilometers upstream of Montreal City where it appears there still existed a significant amount
of forested and unsettled riverfront. Instead, it seems these people chose to settle inwards, i.e.
density was building upon itself. Again, the PRDH database might elucidate this story more by
giving us an understanding of the geographic orientation of kinship networks.
118
Figure 68: Kernel Density Analysis of Dwellings. This method adds together the inverted
distances to neighboring dwellings (this weights closer neighbors more heavily than distant ones)
within a 5km radius (the distance that most healthy people can walk in about an hour).
119
Figure 69: Kernel Density Analysis of Dwellings in Montreal. Light grey indicates cleared land, dark grey forest or swamp.
120
So while one major pole of the colony’s population around Quebec City and environs
was rooted in a seventeenth-century imperial political agenda, the other in the government of
Montreal emerged in the eighteenth century more organically out of the particularities of
environment there. Even as these outlier dwellings in Montreal were situated distant from major
waterways, they were attracted to the social density made possible by the convergence of those
waterways.
This analysis is, again, preliminary and suggestive. It could be repeated to think about
other kinds of proximity—to churches, mills, or roads as a few examples. And we might draw in
other sources bases such as the PRDH or other historical maps. This could have a chronological
orientation—i.e. looking at earlier or later historical maps depicting the same terrain to ask
questions about change. Here the 1685—6 and 1709 surveys offer tantalizing possibilities
because they both record people’s names.
We might also think about broadening our geographic frame. The Murray Atlas’s
enthusiastic reception in the halls of power helped launch many more geographic surveys
throughout British North America over the next two decades.
105
These later cartographic projects
were better documented and executed with better tools—which likely means they had better
positional accuracy than the Murray Atlas. Subjecting the Murray Atlas to additional tests of
completeness may suggest that it will be worthwhile to digitally process these other snapshots of
British North America in the era of imperial crisis and colonial revolt.
Such a dataset stretched across colonies along the eastern seaboard should illustrate some
of the typical spatial models by which Europeans colonized North America while helping us
determine what made the outliers, outliers. In other colonies, for example, do we also find that
Europeans stay quite so close to waterways? Or do we find that different environments, climates,
105
Hornsby, Surveyors of Empire; Jones, “Charting the Imperial Will.”
121
economic systems, legal practices, political structures, ethnic and racial compositions, gender
ratios, and/or religious sects created entirely different settlement patterns? Such a project will
also allow us to revise demographic estimates that have to date been based in what were likely
flawed censuses.
Whatever questions scholars might want to ask, they now have a historical geodatabase to
help them. They also have a greater sense of the reliability or unreliability of three important pre-
1800 quantitative source bases (maps, parish registers, and censuses). Most important, however,
this project has developed and documented a robust methodology for using historical maps to
create spatial data and then evaluating that spatial data. This process involves conceptual labor in
conjunction with historiographical and historical research. It should be rooted in the archives but
not limited by them. Researchers must prepared to expand time and energy to create the vector
data and then, to spatially adjust them. They should look to create spatial and temporal alignment
with other records. And they should be comfortable with and able to document their uncertainty.
The result of this methodology yields geospatial data, yes. But it also provides a deeper
appreciation of the historical map and the geography it represents.
122
References
Anderson, Fred. 2000. Crucible of War: The Seven Years’ War and the Fate of Empire in British
North America, 1754-1766. New York: Alfred A. Knopf.
Barber, Peter. 1990. “Necessary and Ornamental: Map Use in England Under the Later Stuarts,
1660-1714.” Eighteenth-Century Life 14 (3):1–28.
Barber, Peter, and A. Stuart Mason. 1991. “'Captain Thomas, the French Engineer’: And the
Teaching of Vauban to the English.” Proceedings of the Huguenot Society 25:279–87.
Black, Fiona, Bertrum MacDonald, and Black, Malcolm. 1998. “Geographic Information
Systems: A New Research Method for Book History.” Book History 1:11–31.
Bolstad, Paul, P. Gessler, and T. M. Lillesand. 1990. “Positional Uncertainty in Manually
Digitized Map Data.” International Journal of Geographical Information Systems, no.
4:399–412.
Bonnel, Jennifer, and Marcel Fortin, eds. 2014. Historical GIS Research in Canada. Calgary:
Univ. of Calgary Press.
Borges, Jorge Luis. 1998. “On Exactitude in Science.” In Jorge Luis Borges: Collected Fictions,
translated by Andrew Hurley, 325. New York, N.Y.: Penguin Books.
Boudreau, Claude, Serge Courville, and Normand Séguin. 1997. Le Territoire: Atlas Historique
Du Québec. Québec : Sainte-Foy [Québec]: Archives nationales du Québec ; Presses de
l’Université Laval.
Branch, Jordan. 2015. The Cartographic State: Maps, Territory and the Origins of Sovereignty.
Cambridge: Cambridge University Press.
Brotton, Jerry. 2013. A History of the World in 12 Maps. New York: Viking Adult.
Censuses of Canada, 1665 to 1871. 1876. Vol. IV. Statistics of Canada. Ottawa: I. B. Taylor.
Charbonneau, Hubert, and Jacques Légaré. 1967. “La Population Du Canada Aux Recensements
de 1666 et 1667.” Population (French Edition) 22 (6):1031–54.
Chiang, Yao-Yi, Leyk Sefan, and Craig Knoblock. 2014. “A Survey of Digital Map Processing
Techniques.” ACM Computing Surveys 47 (1):1–44.
Choquette, Leslie. 1997. Frenchmen into Peasants: Modernity and Tradition in the Peopling of
French Canada. Cambridge, Mass: Harvard University Press.
Conzen, Michael P., ed. 2010. The Making of the American Landscape. 2nd ed. New York:
Routledge.
123
Coppock, J. T. 1968. “Maps as Sources for the Study of Land Use in the Past.” Imago Mundi
22:37–49.
Courville, Serge, and Richard Howard. 2008. Quebec: A Historical Geography. Vancouver:
UBC Press.
Crean, J. F. 1962. “Hats and the Fur Trade.” The Canadian Journal of Economics and Political
Science / Revue Canadienne d’Economique et de Science Politique 28 (3):373–86.
Crespo Solana, Ana, ed. 2014. Spatio-Temporal Narratives: Historical GIS and the Study of
Global Trading Networks (1500-1800). Newcastle upon Tyne: Cambridge Scholars
Publishing.
Donahue, Brian. 2004. The Great Meadow: Farmers and the Land in Colonial Concord. New
Haven: Yale University Press.
Drennan, Robert D., C. Adam Berrey, and Christian E. Peterson. 2015. Regional Settlement
Demography in Archaeology. Clinton Corners, New York: Eliot Werner Publications,
Inc.
Eccles, W. J. 1969. The Canadian Frontier, 1534-1760. New York: Holt, Rinehart and Winston.
Edelson, S. Max. 2017. The New Map of Empire: How Britain Imagined America Before
Independence. Cambridge, Massachusetts: Harvard University Press.
Fischer, David Hackett. 2008. Champlain’s Dream. New York: Simon & Schuster.
Foucault, Michel. 1995. Discipline and Punish: The Birth of the Prison. 2nd Vintage Books ed.
New York: Vintage Books.
Glass, D. V. 1973. Numbering the People: The Eighteenth-Century Population Controversy and
the Development of Census and Vital Statistics in Britain. Farnborough: D. C. Heath.
Gliserman, Nicholas. 2016. “Landscapes of Conflict: Cartography and Empire in Northeastern
America, 1680-1713.” Dissertation, Los Angeles: University of Southern California.
Greene, Evarts, and Virginia Harrington. 1993. American Population Before the Federal Census
of 1790. Baltimore: Genealogical Publishing, Inc.
Gregory, Ian, and Paul S Ell. 2007. Historical GIS: Technologies, Methodologies, and
Scholarship. Cambridge: Cambridge University Press.
Gregory, Ian N, and Alistair Geddes. 2014. Toward Spatial Humanities: Historical GIS and
Spatial History. Bloomington: Indiana University Press.
Habermas, Jurgen. 1962. The Structural Transformation of the Public Sphere: An Inquiry into a
Category of Bourgeois Society. Translated by Thomas Burger. Cambridge, MA: MIT
Press.
124
Harley, J. B. 1988. “Silences and Secrecy: The Hidden Agenda of Cartography in Early Modern
Europe.” Imago Mundi 40 (January):57–76.
Harley, J. B., Barbara Bartz Petchenik, and Lawrence W. Towner. 1978. Mapping the American
Revolutionary War. The Kenneth Nebenzahl, Jr., Lectures in the History of Cartography
at the Newberry Library. Chicago: University of Chicago Press.
Harley, J. B., and David Woodward, eds. 1987. The History of Cartography. 6 vols. Chicago;
London: Univ. of Chicago Press.
Hornsby, Stephen. 2011. Surveyors of Empire: Samuel Holland, J. W. F. Des Barres, and the
Making of the Atlantic Neptune. Montreal: Mcgill Queens University Press.
Hudson, Pat. 2000. History by Numbers an Introduction to Quantitative Approaches. London:
Arnold.
Jacobs, Wilbur R., ed. 1960. Letters of Francis Parkman. 2 vols. Norman: University of
Oklahoma Press.
Johnson, Alexander James Cook. 2011. “Charting the Imperial Will Colonial Administration &
the General Survey of British North America 1764-1775.” Dissertation, Exeter:
University of Exeter.
Knowles, Anne Kelly, ed. 2008. Placing History: How Maps, Spatial Data, and GIS Are
Changing Historical Scholarship. 1st ed. Redlands, Calif: ESRI Press.
Konvitz, Josef W. 1987. Cartography in France, 1660-1848: Science, Engineering, and
Statecraft. Chicago: University of Chicago Press.
Légaré, Jacques. 1988. “A Population Register for Canada under the French Regime: Context,
Scope, Content and Applications.” Canadian Studies in Population 15 (1):1–16.
Levin, Noam, Ruth Kark, and Emir Galilee. 2010. “Maps and the Settlement of Southern
Palestine, 1799-1948: An Historical/GIS Analysis.” Journal of Historical Geography
36:1–18.
Litalien, Raymonde, Denis Vaugeois, and Jean-François Palomino. 2007. Mapping a Continent:
Historical Atlas of North America, 1492-1814. Sillery, Québec: Les éditions du
Septentrion.
Lloyd, Christopher D., and Keith D. Lilley. 2009. “Cartographic Veracity in Medieval Mapping:
Analyzing Geographical Variation in the Gough Map of Great Britain.” Annals of the
Association of American Geographers 99 (1):27–48.
Mahon, Reginald Henry. 1921. Life of General the Hon. James Murray: A Builder of Canada.
London: John Murray.
125
Mancall, Peter C. 2013. “The Raw and the Cold: Five English Sailors in Sixteenth-Century
Nunavut.” The William and Mary Quarterly 70 (1):3–40.
Marshall, Douglas W. 1976. “The British Military Engineers 1741-1783: A Study of
Organization, Social Origin, and Cartography.” Dissertation, Ann Arbor: University of
Michigan.
———. 1980. “Military Maps of the Eighteenth-Century and the Tower of London Drawing
Room.” Imago Mundi 32 (January):21–44.
McNeese, Tim. 2005. The St. Lawrence River. Rivers in World History. Philadelphia: Chelsea
House Publishers.
Meinig, D. W. 1986. The Shaping of America: A Geographical Perspective on 500 Years of
History. New Haven: Yale University Press.
Moretti, Franco. 1998. “Atlas of the European Novel, 1800-1900.” London; New York: Verso.
———. 2007. Graphs, Maps, Trees: Abstract Models for Literary History. London; New York:
Verso.
Parkman, Francis. 1877. Count Frontenac and New France under Louis XIV. Boston: Little,
Brown and Company.
Pearson, Brooks. 2005. “Comparative Accuracy in Four Civil War Maps of the Shenandoah
Valley: A GIS Analysis.” The Professional Geographer 57 (3):376–94.
Pedley, Mary Sponberg. 2005. The Commerce of Cartography: Making and Marketing Maps in
Eighteenth-Century France and England. The Kenneth Nebenzahl, Jr., Lectures in the
History of Cartography. Chicago: University of Chicago Press.
Rivers of North America. 2005. Boston: Elsevier/Academic Press.
Scull, G. D., ed. 1881. The Montresor Journals. New York: Printed for the New-York Historical
Society.
Shields, James. 1980. “The Murray Map Cartographically Considered: A Study of General
James Murray’s Survey of the St. Lawrence River Valley in 1761.” Dissertation,
Kingston, Ontario, Canada: Queen’s University.
Shipton, Nathaniel. 1967. “General James Murray’s Map of the St Lawrence.” Cartographica 4
(2):93–101.
Siebert, Loren J. 2000. “Using GIS to Document, Visualize, and Interpret Tokyo’s Spatial
History.” Social Science History 24 (3):537–74.
Soll, Jacob. 2011. The Information Master: Jean-Baptiste Colbert’s Secret State Intelligence
System. Ann Arbor: University of Michigan Press.
126
Stevenson, Edward Luther. 1906. Maps Illustrating Early Discovery and Exploration in
America, 1502-1530. New Brunswick, NJ.
Talbert, Richard, and Tom Elliot. 2008. “Richard J. A. Talbert and Tom Elliott, “New Windows
on the Peutinger Map of the Roman World.” In Placing History: How Maps, Spatial
Data, and GIS Are Changing Historical Scholarship, edited by Anne Kelly Knowles, 1st
ed, 199–218. Redlands, Calif: ESRI Press.
Tobler, W. R. 1966. “Medieval Distortions: The Projections of Ancient Maps.” Annals of the
Association of American Geographers 56 (2):351–60.
Tuan, Yi-Fu. 1977. Space and Place: The Perspective of Experience. Minneapolis: University of
Minnesota Press.
Wells, Robert. 1975. Population of the British Colonies in America Before 1776. Princeton:
Princeton University Press.
Willigan, J. Dennis, and Katherine A. Lynch. 1982. Sources and Methods of Historical
Demography. Studies in Social Discontinuity. New York, N.Y: Academic Press.
127
Appendix A: Models for Reconstructing Missing Parish Records
Reconstructing Canadian population using the parish registers depends upon successfully
estimating the number of births and deaths for the years and parishes with missing records. I
embarked on this task with multiple models. To gage their reliability, I first used them to
estimate values I already knew—in other words to gauge whether they predicted those values
with any precision.
This approach hinges upon the assumption that the missing data and the surviving data
are similar to each other. This assumption is partially flawed. Some data are missing for what
appear to be random reasons. But poor record keeping during the earliest years of a parish was a
recurring pattern, suggesting a potential correlation between the cause of the absent data and the
value of that data. Because we only have a handful of parishes with complete records in their
early years, we have less insight into the overall nature of this kind of data. I summarize my
models in Table 8 and explain them in more detail throughout the next few paragraphs.
Following this discussion, I present the error for each of these models in
Table 9. I have cartographically visualized the reconstructed natural population growth in
Figure 70. Finally, the population model I constructed for the entire colony with this
reconstructed data follows in Appendix B.
The first two models (Model 1 and 2) averaged together adjacent years. Model 1
averaged the year directly before and following; Model 2 drew upon the three years directly
before and following. Missing records were not included in the average. For example, if we were
trying to estimate the value for one year and the previous year was missing, then Model 1 would
replicate the value of the following year. These models preclude the possibility of a number
above or below the range of adjacent years, thereby eliminating the possibility of outliers. In
128
other words, they weed out unusual numbers that don’t fit within the range of the parish but are
slightly problematic for years where war, disease, famine, or unusual bounty.
Table 8: Description of Models to Estimate Missing Parish Values
Model Description
1 Averages the values for the previous and subsequent year
2 Averages the values for the previous and subsequent three years
3
Multiplies value for previous year by the average growth rate for the year
(based on sum of all parishes with complete records)
4
Divides value for next year by the average growth rate for the year (based on
sum of all parishes with complete records)
5
Multiplies value for previous year by the average growth rate in the region for
the year (based on sum of all parishes with complete records)
6
Divides value for next year by the average growth rate in the region for the
year (based on sum of all parishes with complete records)
3F
Method 3, Excludes Values that Fall Beyond the Range of Known Values for
the Parish during its Lifetime
4F
Method 4, Excludes Values that Fall Beyond the Range of Known Values for
the Parish during its Lifetime
5F
Method 5, Excludes Values that Fall Beyond the Range of Known Values for
the Parish during its Lifetime
6F
Method 6, Excludes Values that Fall Beyond the Range of Known Values for
the Parish during its Lifetime
A(verage)
Average of Methods 1, 2, 3F, 4F, 5F, 6F (When values available). If the value
falls below 0 (i.e. for births), given default value of 0.
7
Uses Ratio of Known Data from the Parish for the same Years for the Region,
Extrapolates data for missing years
8
Starts Parishes with Complete Records at the Same Time, Averaging
Together Values for Y1, Y2, Y3, etc.
Models 3, 4, 5, and 6 employed growth rates to calculate either forward from the year
preceding (Models 3 and 5) or backwards from the year following (Models 4 and 6). I derived
these growth rates from parishes with complete records during the relevant two-year period both
throughout the colony (Models 3 and 4) and region (Models 5 and 6). In other words, if was
trying to determine growth rate to use to calculate forward from 1750 or backward from 1751, I
would only include parishes for which I had records for both years (1750 and 1751).
129
Certain problems presented themselves for Models 3 to 6. First, they could produce
results that fell beyond the range of recorded values for the parish. While it is certainly possible
that the missing parish registers included outliers, I wanted to see what the results looked like
without them as well. A second problem arose: A large rate of change (in the region or
throughout the colony) in a given year could create negative results in the cases of births or
deaths. This is not necessarily a problem for predicting natural growth, which is negative when
deaths outweigh births. But for this study, I have decided to filter out the unborn and undead.
These two layers of filtering (denoted in the model name with an “F” following the number)
yielded only marginally lower overall rates of error but they improved the distribution of that
error—so, for example, we saw fewer individual cases of error greater than 100%.
At a certain level, Models 3 through 6 add a certain kind of nuance to Models 1 and 2 but
none will help us find the value of missing data if we don’t have records for adjacent years. Said
differently, however well these models worked for estimating randomly missing records, they
did much less to establish the values for records missing in the early years of a parish—
especially when the run of missing years lasted for a decade or longer. In these cases, I
developed an additional two methods for estimating values. Model 7 determined the ratio of
values between parish and region for the years in which parish data is known, using that ratio to
extrapolate the data for missing years. Model 8 drew upon the parishes with complete records to
create average values for the first, second, third, etc. years.
The problems with these models should be somewhat obvious. Model 7 does not factor in
other missing parish registers in calculating its ratio or extrapolating backwards from it. Nor does
it give us the values for the first years of the first parish operating in a region (if the parish
register is missing, without a second parish, there is no value for the region). Therefore this
130
model will work best during the later years when more parishes populate a region—precisely
when we need it least. Model 8 assumes that all parishes are essentially alike—this seems quite
dubious. Not surprisingly, the error rate was high for both models (e.g. these methods
underestimated births by 55% and 35% respectively), reflecting highly flawed assumptions.
Probably the error was actually much higher for Model 8 because it drew upon biased aggregate
data. Said differently, the only way we would know the error rate for a given parish was if that
parish had data that survived—and if it had data that survived, it was guiding our estimate in the
first place. The best we can say here is, first, that something might be slightly better than
nothing. Second, it seems that these methods undercount rather than over count so that we find
an answer slightly closer to the truth but without overshooting. Third, we can interrogate our
results using CBR, which we do in the next appendix.
The final model, which I have named with the letter “A” for average, averaged together
Models 1, 2, 3F, 4F, 5F, and 6F. If a particular model did not generate an estimate for a given
parish in a given year (e.g. if Model 1 had no data from the adjacent year before and after) it was
not included in the average. I experimented with a few other averages but found this one yielded
the most impressive results: A low overall (and generally well-distributed) rate of error. As
measured against known data, Model A over counted natural growth (births minus deaths) by
0.29% and undercounted births by 0.87%. See
Table 9 below for the error rates of all the different models.
131
Table 9: Model Error for Estimating the Values of Missing Parish Registers
Percent Error Estimating:
Model Births Natural Population Growth
1
1.18% 1.95%
2
0.45% 0.94%
3
1.46% 7.07%
4
-3.23% 1.91%
5
3.34% 2.52%
6
8.80% 9.11%
3F
0.83% 6.51%
4F
-1.41% 4.80%
5F
4.70% 5.00%
6F
0.27% 8.96%
A(verage)
-0.87% 0.29%
7
-55.09% -8.28%
8
-35.66% 41.14%
After testing the error rate of these models, I followed a hierarchy in actually estimating
the missing values. First, I filled in the missing records with results from Model A. For years
where this did not yield an estimate, I used the results from Model 7 (even though this had a
higher rate of error than Model 8, at least I knew I could trust the estimate of error). Only for the
values that were still missing at this point did I use Model 8. For a cartographic visualization of
these results see Figure 70. One final note is that we should be able to improve upon these results
in the future with access to the full PRDH database: Because vital records often reference earlier
life events (for which the records might not survive), we should be able to fill in more missing
data (e.g. marriage certificates discussing when and where the husband and wife were born).
132
Figure 70: Regional Distribution of Projected Natural Population Growth (Raw Numbers) by
Decade (1620-1760). White coloring indicates no parishes yet established in the region.
133
Appendix B: Modeling Population in Canada with Parish Registers
Having created estimates for the value of missing data (Appendix A), my next objective was
creating population estimates for the colony as a whole from its founding (1616) to the date of
the last census considered for this study (1765). Doing this for the whole colony is
straightforward enough: For a given year, we add A) natural growth (births minus deaths), B) net
immigration to the colony, and C) the population from the previous year. Fortunately, Leslie
Choquette has reconstructed the numbers for net migration to Canada.
106
For natural growth, I
used two sets of numbers: The first as recorded only in the parish registers that survive; and the
second including the estimates for missing records. This illustrates the estimated effect of the
missing data. The results are below in
Table 10, Figure 71, and Figure 72.
Table 10: Population Estimates and Source Comparisons
Year
Population from Source
Net
Migration
to Canada
Surviving
registers
Registers with
estimates for
missing data
Census
(# of
People)
Census
(# of
Families)
Map/Census
(# of
Dwellings)
1620 22 22 36
N/A
1640 958 1,325 1,470
1660 4,983 7,097 7,340
1680 9,591 17,350 18,308
1700 11,500 27,226 29,343
1720 12,530 39,985 42,768
1740 15,126 63,653 66,931
1760 23,546 93,929 97,571 11,043 12,911 (Map)
1762 23,666 98,111 101,782 66,068 N/A N/A
1765 23,846 105,145 108,950 55,100 10,660 9,930 (Census)
106
Choquette, Frenchmen into Peasants, 162.
134
Figure 71: Canadian Population, 1615-1765
Figure 72: Share of Canadian Population from Natural Growth and Immigration, 1615-1765
What happens if we subject these estimates to the same CBR test that we employed to the
census and Murray Atlas? In the earliest years of the colony (roughly to 1640), the CBR looks
erratic and at times impossibly high. The gender ratio of the early colony skewed heavily male,
so we should expect a very low CBR. Yet the model estimating values for missing parish data
does a poor job in these early years (Figure 73)—whether because it underestimates overall
135
population or overestimates the number of births—providing a CBR in 1626 that almost reaches
250. This is not entirely surprising because these estimates come from what was the least reliable
model, as discussed in Appendix A. The small number of operational parishes during this period
would, moreover, weight these bad estimates more heavily. The numbers for only the surviving
registers also look unusual, jumping back and forth to and from zero, peaking above one hundred
in the late 1630s. In some sense, the poor results here are not too surprising. Everything here is
complicated by large gaps in record keeping and the erratic nature of a small early colonial
population.
Figure 73: CBR Calculated from Only Surviving Registers vs. Surviving Registers with
Estimated Values for Non-Surviving Registers, 1616-1651.
What we can say about this population projection is that after about the year 1645, it
provides a CBR within reasonable bounds that broadly conforms to what we might expect. We
see, for example, a low CBR in the 1650s that increases slowly in the 1660s and then more
dramatically in the 1670s, reflecting the state-sponsored migration of young (unmarried) women
to the colony in the 1660s to balance the heavily male gender ratio. The CBR calculated from
surviving registers generates a trendline with a higher R
2
value (0.71299) than does the one using
our estimated values (R
2
=0.56874). We might not necessarily read too much into this, as it might
136
mean that including estimates merely exaggerates the seventeenth-century volatility we already
see in the records. Whichever population estimate derived from PRDH data we use, we can see
that it points to a much larger population than either 1762 or 1765 census (
Table 10). That the natural growth alone is greater than the census estimates gives us a
strong clue as to how much the censuses were undercounting people.
Figure 74: CBR Calculated from Surviving Registers Only vs. Surviving Registers and
Estimated Values for Non-Surviving Registers, 1645-1765.
137
Appendix C: Comparing 1760 and 1765 Censuses by Region
Region Families 1760 Families 1765 Difference (%)
Quebec West 297 465 56.6%
Quebec South 333 381 14.4%
Quebec City and Suburbs* 665* 858 29.0%
Chaudiere-Etchemin River Valley 524 647 23.5%
Isle d'Orleans 383 420 9.7%
Quebec Shore S of I Orleans 689 820 19.0%
Quebec Shore N of I Orleans 286 313 9.4%
Quebec East 155 291 87.7%
Quebec North 161 186 15.5%
Quebec (Unmapped) 556 599 7.7%
TOTAL QUEBEC 4049 4980 23.0%
Trois Rivieres South 415 387 -6.7%
Trois Rivieres West 326 424 30.1%
Trois Rivieres East 164 142 -13.4%
Trois Rivieres North 310 307 -1.0%
Trois Rivieres (Unmapped) 66 0 N/A
TOTAL TROIS RIVIERIES 1281 1260 -1.6%
Montreal West 102 116 13.7%
Isle Jesus 402 467 16.2%
Montreal Island* 1285* 937 -27.1%
Montreal South Shore 1198 851 -29.0%
Richelieu River Corridor 676 803 18.8%
Montreal North Shore 957 1050 9.7%
Montreal (Unmapped) 194 196 1.0%
TOTAL (MONTREAL) 4814 4420 -8.2%
TOTAL (ALL) 10144 10660 5.1%
Asterisk (*) indicates data has been subtracted to facilitate comparison (i.e. for the parishes with
missing data in the 1765 census, the 1761 parish data has not been incorporated into the number
of families by region to create parity in the comparison. Of course, these comparisons might still
be problematic, if the census takers defined the parish boundaries differently).
138
Appendix D: Changes in Population According to 1762 and 1765 Censuses
Region 1762 Census
107
1765 Census
108
Difference (%)
Quebec West 2538 2496 -1.7%
Quebec South 1902 1923 1.1%
Quebec City and Suburbs 5602* 4531* -19.1%
Chaudiere-Etchemin River Valley 2876 3282 14.1%
Isle d'Orleans 2458 2303 -6.3%
Quebec Shore S of I Orleans 4619 4443 -3.8%
Quebec Shore N of I Orleans 1635 1762 7.8%
Quebec East 1451 1590 9.6%
Quebec North 1032 1054 2.1%
Quebec (Unmapped) 2648 3339 26.1%
TOTAL QUEBEC* 21159 22192 4.9%
Trois Rivieres South 1065 1918 80.1%
Trois Rivieres West 2974 2088 -29.8%
Trois Rivieres East 646 724 12.1%
Trois Rivieres North 1524 1597 4.8%
Trois Rivieres (Unmapped) 283
TOTAL TROIS RIVIERIES 6972 6327 -9.3%
TOTAL (QUEBEC
AND TROIS RIVIERES*)
28131 28519 1.4%
Asterisks (*) indicate the total without data for Quebec City.
107
[Report on the Government of Quebec] (June 5 1762), CO 323/15, TNA; [Report on the Government of Trois
Rivieres] (April 1762), CO 323/16, TNA.
108
Censuses of Canada, 1665 to 1871 (Ottawa: I. B. Taylor, 1876), 62-68.
139
Appendix E: Hypothetical Colony-Wide CBRs for 1760 Census based on Average
Family Size
Region Name Births Families
CBR if Family Size =
4 5 6 7 8
Montreal West 34 102 83.33 66.67 55.56 47.62 41.67
Trois Rivieres South 114 415 68.67 54.94 45.78 39.24 34.34
Quebec West 127 297 106.90 85.52 71.27 61.09 53.45
Trois Rivieres West 165 326 126.53 101.23 84.36 72.30 63.27
Isle Jesus 174 402 108.21 86.57 72.14 61.83 54.10
Montreal Island 816 2117 96.36 77.09 64.24 55.06 48.18
Montreal South
Shore
485 1198 101.21 80.97 67.47 57.83 50.61
Richelieu River
Corridor
317 676 117.23 93.79 78.16 66.99 58.62
Montreal North
Shore
587 957 153.34 122.68 102.23 87.63 76.67
Trois Rivieres East 54 164 82.32 65.85 54.88 47.04 41.16
Quebec South 136 333 102.10 81.68 68.07 58.34 51.05
Quebec City and
Suburbs
601 1548 97.06 77.65 64.71 55.46 48.53
Chaudiere-Etchemin
River Valley
202 524 96.37 77.10 64.25 55.07 48.19
Isle d'Orleans 121 383 78.98 63.19 52.65 45.13 39.49
Quebec Shore S of I
Orleans
354 689 128.45 102.76 85.63 73.40 64.22
Quebec Shore N of I
Orleans
88 286 76.92 61.54 51.28 43.96 38.46
Quebec East 96 155 154.84 123.87 103.23 88.48 77.42
Quebec North 55 161 85.40 68.32 56.94 48.80 42.70
Trois Rivieres North 133 310 107.26 85.81 71.51 61.29 53.63
TOTAL 4659 11043 105.47 84.38 70.32 60.27 52.74
140
Appendix F: Regional Distribution of Dwellings in Murray Atlas (with
Extrapolated Population and CBR) by Region
Region
Dwellings
(#)
Estimated Population
(7 People per
Dwelling)
CBR
Montreal West 259 1838.9 12.5
Trois Rivieres South 480 3408 22.0
Quebec West 573 4068.3 32.7
Trois Rivieres West 587 4167.7 30.7
Isle Jesus 416 2953.6 46.7
Montreal Island 1066 7568.6 99.2
Montreal South Shore 1420 10082 42.6
Richelieu River Corridor 915 6496.5 38.3
Montreal North Shore 1427 10131.7 38.8
Trois Rivieres East 186 1320.6 26.5
Quebec South 611 4338.1 21.6
Quebec City and Suburbs 1905 13525.5 41.5
Chaudiere-Etchemin River
Valley
540 3834 34.4
Isle d'Orleans 330 2343 28.6
Quebec Shore S of I Orleans 1028 7298.8 31.2
Quebec Shore N of I Orleans 286 2030.6 35.5
Quebec East 158 1121.8 60.6
Quebec North 132 937.2 44.8
Trois Rivieres North 592 4203.2 24.5
Abstract (if available)
Abstract
This project employs a mix of archival and digital techniques to evaluate an extensive geographic survey of the St. Lawrence River valley, conducted under auspices of the Quebec Governor James Murray following the 1760 British conquest of Canada. I show how scholars have misread the historical evidence surrounding the production and dissemination of this survey by ignoring the patronage motives of those involved in its production. I also use archival evidence to examine contemporary practices and ideas surrounding maps and accuracy. I discuss how I built a Historical Geographic Information System by manually creating and spatially adjusting vector data from Murray’s personal copy of the survey (which was composed of forty-four individual map sheets measuring forty-five by thirty-six feet when fully assembled). This dataset allowed me to establish that the survey demonstrated a high degree of spatial accuracy for the eighteenth century. Here I also discuss methods for retroactively creating administrative boundaries, which were undocumented in the period. This allowed me to create spatial interoperability with other contemporary quantitative records—historical censuses and parish registers, which recorded births, marriages, and deaths—to evaluate the reliability of these various administrative technologies of state during the early modern period. I conclude that while the historical censuses undercounted people, the survey and parish registers support each other’s conclusions, which suggest their demographic accuracy. This work serves as a proof of concept for a much larger spatial humanities project that would employ these same techniques to digitally process the series of other geographic surveys conducted throughout British North America between 1765 and 1777 to capture a geographic snapshot of the late colonial period.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Validation of volunteered geographic information quality components for incidents of law enforcement use of force
PDF
The role of precision in spatial narratives: using a modified discourse quality index to measure the quality of deliberative spatial data
PDF
Walking to the Longhouse: a deep map of the Central New York Military Tract & its indigenous history
PDF
Enriching the Demographic Survey sampling for the Los Angeles County Annual Homeless Count with spatial statistics
PDF
Residential housing code violation prediction: a study in Victorville, CA using geographically weighted logistic regression
PDF
Implementing spatial thinking with Web GIS in the non-profit sector: a case study of ArcGIS Online in the Pacific Symphony
PDF
Finding environmental opportunities for early sea crossings: an agent-based model of Middle to Late Pleistocene Mediterranean coastal migration
PDF
A spatial analysis of veteran healthcare accessibility
PDF
Social media to locate urban displacement: assessing the risk of displacement using volunteered geographic information in the city of Los Angeles
PDF
Peoples of Washington historical geographic information system: geocoding culture using archival standards
PDF
Landscapes of conflict: cartography and empire in northeastern America, 1680-1713
PDF
Mapping punk music and its relative subgenres
PDF
Is the likelihood of waterfowl presence greater on conserved lands?
PDF
Demonstrating GIS spatial analysis techniques in a prehistoric mortuary analysis: a case study in the Napa Valley, California
PDF
The geography of voter power in the U.S. electoral college from 1900-2012
PDF
A unified geodatabase design for sinkhole inventories in the United States
PDF
Human suffering during wartime: a StoryMap of violations of international law during the Russo-Ukrainian War
PDF
Surface representations of rainfall at small extents: a study of rainfall mapping based on volunteered geographic information in Kona, Hawaii
PDF
That sinking feeling: predicting land subsidence in California’s San Joaquin Valley with a spatial regression model
PDF
Effect of spatial patterns on sampling design performance in a vegetation map accuracy assessment
Asset Metadata
Creator
Gliserman, Nicholas
(author)
Core Title
Assessing the reliability of the 1760 British geographical survey of the St. Lawrence River Valley
School
College of Letters, Arts and Sciences
Degree
Master of Science
Degree Program
Geographic Information Science and Technology
Publication Date
01/18/2018
Defense Date
12/11/2017
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
British military cartography,Canada,geographic survey,GIS,historical demography,History,history of cartography,James Murray,Murray Atlas,Murray Map,North America,OAI-PMH Harvest,spatial humanities,St. Lawrence River
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Kemp, Karen (
committee chair
), Sedano, Elisabeth (
committee member
), Vos, Robert (
committee member
)
Creator Email
gliserma@usc.edu,Nicholas.Gliserman@Gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c40-463472
Unique identifier
UC11266362
Identifier
etd-GlisermanN-5958.pdf (filename),usctheses-c40-463472 (legacy record id)
Legacy Identifier
etd-GlisermanN-5958.pdf
Dmrecord
463472
Document Type
Thesis
Rights
Gliserman, Nicholas
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
British military cartography
geographic survey
GIS
historical demography
history of cartography
James Murray
Murray Atlas
Murray Map
spatial humanities