Académique Documents
Professionnel Documents
Culture Documents
P A T R I C I A T H O R N T O N * AND S H E R R Y O L S O N
With respect to infant survival over the forty years 18601900, did
Montreal s three largest cultural communities French Canadian, Irish
Catholic, and Anglo-Protestant respond in the same way to a period of
intense industrialization ? In an examination of three birth cohorts, we
anticipated finding higher mortality in working-class families who bore
the brunt of industrialization, and, since the city was growing larger,
dirtier, and more densely populated, we anticipated a trend to stronger
expression of environmental effects. In the years 18601880 the population
doubled in the urbanized area, and it doubled again in 18801900. Urban
growth, by giving value to land, induced intensification of its use, achieved
by construction of triplex (triple-decker) row housing in factory suburbs.
Although residential densities did not approach those of tenement-house
districts of New York or Chicago, peak densities doubled between 1880
and 1900, reaching levels not seen before in Montreal."
We did not find a straightforward relationship with income or social
status, but present here the evidence for a powerful cultural dimension
which persisted over three generations. We propose an interpretation in
95
.
For infants born in Montreal and its suburbs 1 January through 31
December 1859, 1 April 1879 through 31 March 1880, and 1 April 1899
through 31 March 1901 (a two-year span hereafter referred to as the
cohort of 1900), we collected records of baptisms and burials, and
matched them with the surviving children or families recorded in the next
census (January 1861, April 1881, or April 1901). We also employed city
directories (compiled annually in June), the municipal taxroll (compiled
annually in May), and subsequent baptismal or burial records, to ensure
that all families retained in the samples were under observation for 12
months after the birth. In the parish records a search for next births was
designed to cover a span of 48 months after the target births.
Quebec parish records rival those in Europe for their completeness and
accuracy. While record-matching is as demanding as elsewhere, am-
biguous cases (excluded) are rare since the Catholic records contain a
large number of variables, including mothers maiden name and the
child s age at death. Protestant church records are more variable in the
information they supply, but matching of Anglo-Protestant records is
favoured by a greater diversity of surnames and more reliable appearance
in city directories.#
For the 1900 cohort, on which we focus in this article, records were
collected from 27 Catholic and 100 Protestant churches, and compiled to
ensure the closest possible comparison with findings we have reported for
cohorts of 1860 and 1880. Unless otherwise specified, the tables and
figures refer to a sample of 2363 infants, all of whom, except for 23 sets
of twins,$ were born to separate families over the two-year period. As
compared with the earlier cohorts, the span over which we tracked next
96
births is longer (48 months rather than 33 months), and our search of the
census is more thorough.% The census of 1901 is the first which consistently
records street addresses, and these can be matched with addresses
reported in cemetery registers.& For the five years beginning 1 April 1899
we collected deaths of married men and women under 50. Each set of
records was compiled independent of the others, except for a search for
date of marriage of parent couples in genealogical indexes.
The consistent coverage of Montreal sources would permit a wide
variety of sampling strategies. Experience with the first cohort convinced
us of the value of reducing sample size in order to enlarge the array of
variables and permit confrontation of sources. In each successive exercise
we reduced the density of sampling : in 1860 100 per cent of births, in 1880
50 per cent, in 1900 12 per cent. The 1900 sample was drawn by selection
of surnames beginning with the letters B and R.' One can go too far, of
course, and analysis of the 1900 cohort is constrained by the small size of
Protestant and Irish samples (n l 285 and 270), a flaw in design, as we did
not anticipate so radical a reduction of fertility (discussed below).
We explored the potential impact on infant survival of the variables
listed in Table 1, as indicators of the environment surrounding the child.
From the birth record we established sex, twin births, and the cultural
identity.( Municipal rental valuations (for the taxe locatie) are an
excellent approximation of market rents and therefore a measure of
household purchasing power. From the same source we derived categories
of occupational status for the fathers (according to ranks of median rent
of all householders of the same occupation), an estimator for local
residential density (households per kilometre of street frontage), and an
indicator of the local social environment : street rent (that is, median rent
in the street segment).) The census, in addition to mothers age, provided
numbers in the household and rooms in the dwelling, hence an estimator
for crowding : the ratio of persons to rooms. Since segregation along
linguistic lines was considerable in 1900, the child is also situated in a
residential district characterized by the percentage of French-speaking
households.
For all three cultural communities, addresses were established for 95 per
cent, a rental value (from the taxroll) for 73 per cent, and a census match
for three-quarters (81 per cent of French Canadians). For multivariate
analyses which require reference to both census and taxroll, sample sizes
are, in all three groups, reduced to two-thirds.
The representativity of the sample can be appraised against a
corresponding sample of all households in the taxroll. We had used the
entire rental taxroll of May 1901 (65,000 households) to create a scale of
occupational status, to estimate residential densities and median rents for
97
T 1
Infant mortality and sample sizes by cultural community and selected
independent ariables
Sex
Male 902 51.9 25.6 142 54.8 18.3 137 52.1 13.9
Female 835 48.1 20.6 117 45.2 12.8 126 47.9 15.9
Twin births
Singleton 1704 98.1 22.8 253 97.7 259 98.5
Twin 33 1.9 45.5 6 2.3 4 1.5
Household rent
High 208 15.6 23.6 57 29.5 7.0 77 39.7 13.0
Moderate 574 43.1 23.1 84 43.5 17.9 94 48.5 11.7
Low 550 41.3 22.8 52 26.9 19.2 23 11.9 17.4
Occupational status
High 276 16.0 22.8 57 22.1 17.5 75 29.4 14.7
Moderate 1018 58.9 24.0 150 58.1 14.7 147 57.6 12.9
Low 435 25.2 22.1 51 19.8 17.6 33 12.9 24.2
Density in street segment
High ( 350\km) 623 37.0 24.2 54 22.0 13.0 36 14.5 13.9
Mod. (226350\km) 535 31.8 25.8 95 38.6 24.2 66 26.5 15.2
Low ( 225) 526 31.2 21.1 97 39.4 11.3 147 59.0 14.3
Rent in street
High ( $120) 169 10.3 23.7 64 26.3 7.8 86 36.9 14.0
Mod. ($61$119) 695 42.2 23.9 119 49.0 20.2 108 46.4 13.9
Low ($60 or less) 783 47.5 23.9 60 24.7 18.3 39 16.7 15.4
Room-crowding
2 per room 231 16.9 19.9 24 12.0 10 5.1
2 per room 1133 83.1 22.2 176 88.0 187 94.9
District (% French)
70 % 1366 82.6 24.8 111 45.7 19.8 81 34.0 13.6
4069 % 225 13.6 19.6 90 37.0 16.7 56 23.5 21.4
40 % 62 3.8 17.7 42 17.3 7.1 101 42.4 10.9
a Col. l column.
b Rate l deaths on days 2364 per 100 first-day survivors.
98
99
T 2
Strength of relationship between language group and other ariables,
1900
a Df l degree of freedom.
b P l level of significance.
c Significant to 5 % level or better.
100
T 3
Strength of relationship between surial and other ariables, by language
group, 1900a
Anglophone Francophone
F 1 . Infant mortality rates (excluding first-day deaths), 1860, 1880 and 1900. All
rates are lower for 1880, but the three cohorts show a consistent pattern of higher mortality
among French Canadian infants (F), most apparent after the first month. The other cultural
communities identified are Irish Catholic (I) and Anglo-Protestant (P).
since Irish Catholic families showed the lowest profiles of rent and fathers
occupation, yet their infants survived as well as those in the Anglo-
Protestant sample with its high rents and higher occupational profile.
In dealing with each successive cohort, we reviewed the methodology,
experimented with alternative measures of status and conceivable
interaction effects. For the cohort of 1900, as well, we cannot explain
101
F 2 . Occupational status and survival, Montreal, 1900. Bars indicate the distinctive
profile of occupational status in each of the three communities : a larger proportion of high-
status occupations (AB) among Protestants, a wide range among the Irish, more semi-skilled
(E) and labourers (F) among French Canadians. Dots show the erratic levels of infant
mortality in the subsamples (2364 days).
102
T 4
Infant mortality in Montreal by cultural community, 1900a
a All these figures exclude 27 infants designated as stillborn : columns All include 34
infants of Jewish and other cultural groups.
infants lived in solidly French neighbourhoods (70100 per cent) with high
mortality rates, but living in a mixed (4069 per cent) or anglophone-
dominated neighbourhood ( 40 per cent French) brought their survival
rates up to the anglophone average. The neighbourhood effect is also
apparent for Irish Catholic infants, who more often lived in mixed
neighbourhoods, but ambiguous for Protestant infants.
In other words, despite evidence from studies elsewhere, Montreal
families in the top-ranking third of the occupational profile were no more
successful than labourers of the same cultural group in protecting their
infants. The results are consistent in the three cohorts (see Figures 1 and
2). If cultural identities and degrees of cultural segregation had a more
powerful impact on an infants chances than material conditions,"' we
must look more closely at what constitutes culture. It is obviously
something richer and more complicated than what is implied in a cross-
tabulation of religion and language. What cultural practices might have
affected infant survival in Montreal ?
Let us simplify the discussion by focusing on the cohort of 1900 (see
Table 4). Of infants who survived the first day, 20.9 per cent died within
the first year, almost identical with the cohort of 1860, and, despite some
success in campaigns against smallpox (to be discussed), infant mortality
remained as threatening in Montreal as 40 years previously (Table 5)."(
Neonatal mortality (days 228), of the order of 3 to 5 per cent except for
the Irish, varied relatively little, but for post-neonatal deaths (days
103
T 5
Comparatie mortality in the three birth cohorts
1860 1880 1900 1860 1880 1900 1860 1880 1900 1860 1880 1900
Post-neonatal (29364 days) 21.2 15.8 19.0 13.7 10.4 13.9 13.2 10.6 11.4 17.7 14.3 17.3
Neonatal (228 days) 4.0 3.6 3.7 3.8 3.3 1.5 3.8 3.7 5.2 3.9 3.7 3.6
Infant (2364 days) 25.2 19.4 22.7 17.5 13.7 15.4 17.0 14.8 16.6 21.1 18.0 20.9
Index of variation
Post-neonatal (29364 days) 100 100 100 65 66 73 62 67 60 100 81 98
Neonatal (228 days) 100 100 100 95 92 41 95 103 141 100 95 92
Infant (2364 days) 100 100 100 69 71 68 67 76 73 100 85 99
T 6
Numbers of deaths by reported causes, 1880 and 1900, days 2364
29364) the gap was very large between the French Canadian and the
other two populations (19.0 per cent versus 13.9 and 11.4), resulting in a
much higher French Canadian infant mortality rate (22.7 per cent versus
15.4 and 16.6).
In order to make sense of the differences, we explored the seasons, ages,
and causes of death. Of 481 infant deaths in the 1900 cohort, a cause was
105
T 7
Causes of death summarized for the three cultural communities, 1880 and
1900
Contagious 39 18 8 3 11 1 58 22
Delivery 29 20 4 1 6 2 39 23
Hydrocephaly 61 13 25 1 24 1 110 15
Intestinal 191 187 43 14 45 9 279 210
Weakness 100 75 15 10 41 11 156 96
Respiratory 40 47 13 5 10 5 63 57
Tubercular 22 29 5 3 4 2 31 34
Other 6 4 2 2 1 2 9 8
Total 488 393 115 39 142 33 745 465
No report 20 11 3 1 11 6 34 18
reported for 96 per cent, and the reported causes were grouped to
highlight environmental factors (see Tables 6 and 7).") As shown in Figure
3, identification of causes had improved, vague and symptomatic rubrics
such as debility, weakness and convulsions forming a smaller proportion
in 1900."* Health Department tabulations over a half-century (see Figure
4) show the important decline of smallpox since 1885, of mortality from
several other childhood diseases after 1900, from diarrhoeal diseases after
1910.#!
Among sample infants of the 1900 cohort, nearly half of infant deaths
after the first day were attributed to intestinal causes such as diarrhoea
and infant cholera (see Table 8), and precisely half of all the deaths
occurred in the summer quarter of the year (June, July, or August). Of
deaths reported as intestinal, 71 per cent occurred in the summer quarter,
thus seasonality is a strong diagnostic of diarrhoeal infections as either a
primary or secondary cause of death. This corresponds to what Woods
and others have described in cities of nineteenth-century Britain.#" In the
Montreal samples, as shown in Table 8, these effects were most
pronounced among French Canadian infants (51.9 per cent of deaths in
summer, 47.6 per cent intestinal) where diarrhoeal mortality amounted to
one-tenth of all births. Deaths of Irish Catholic infants were less often
attributed to intestinal causes (one case in 20 births), but were just as likely
to occur in the summer quarter. For Protestants summer concentration
was almost nil, and only one infant in 30 died of diarrhoeal causes.
106
T 8
Comparatie season and cause of death, 1880 and 1900
Infant deaths (2364 days) 1880 1900 1880 1900 1880 1900 1880 1900
F 3 . Selected causes by age at death, 1880 and 1900 (excludes first-day deaths).
Death records for the two infant cohorts show improvement in identification of causes,
vague and symptomatic rubrics such as teething, weakness, and convulsions forming a
smaller proportion in 1900.
107
108
F 5 . Biometric analysis of infant deaths, 1900. Trend shows the divergence of death
rates for French Canadian infants, with the steeper slopes after three months, a divergence
for Irish Catholics after seven months. The Bourgeois-Pichat formula has been modified to
start from the day after birth.
F 6 . Biometric analysis of infant deaths, 1900 (by season). The steep curve for
summer deaths can be compared with the other nine months of the year, by standardizing
both rates to a 12-month basis.
109
110
T 9
Odds ratios for infant surial in logistic model, 1900
dimensions are independent and cumulative. Once both are taken into
account the cultural identity of the family and the district no other
environmental characteristics show any further impact. The scale at which
neighbourhood is specified seems to be important, since no mortality
differentials can be discerned from properties of the street segment, nor
from properties of the dwelling itself.
Over 40 years there was no apparent improvement in infant survival,
and the differentiating cause was more acute in 1900 : early weaning in an
unsanitary urban environment. In such habitats, historical demographers
have long recognized the protective effects of breastfeeding,#' and scholars
in recent years have given increasing attention to its cultural, social, and
regional variations. A shift in scale of analysis from national aggregates to
localized microstudies has revealed a wide range of factors which
differentiate maternal behaviours in finely textured cultural milieus.#( To
estimate the extent and variations of breastfeeding in Montreal, we turn
to the analysis of birth spacing, which may also be informative with
respect to trends in fertility.
.
For the cohort of 1900, fertility is of considerable interest because these
families were on the cusp of change. In light of McInnis evidence for a
fertility transition apparent by 1891 in Ontario, the multicultural
metropolis is a particular challenge, and Montreal appears on his map as
a fulcrum of change. To obtain nation-wide coverage, McInnis had to rely
on census ratios of young children in households, an indicator whose
adjustment is especially difficult in a population where the cultural mix in
a census enumeration district affected infant deaths as well as births.#) Our
1900 cohort, where birth and death rates can be independently established
for each cultural group, confirms his notion of two cultures of child-
111
T 10
Age at marriage of parents of infants, 1880 and 1900
112
T 11
Median birth interals by cohort and community
Birth interals in months 1860 1880 1900 1860 1880 1900 1860 1880 1900
First child died 1 year 16.9 16.2 16.0 18.6 20.4 16.2 18.4 18.1 19.4
First child survived 1 year 25.0 22.1 21.5 25.8 24.6 23.8 23.5 24.4 25.3
Next born in 18 months (%) 11.4 27.5 33.2 11.2 13.0 21.7 17.6 23.2 19.4
113
F 7 . Birth intervals in French Canadian sample families. When a child died very
young, the mother was likely to have another within 12 to 16 months, as shown above for
French Canadian families in the sample associated with the birth cohort of 1900. Where
infants survived to 12 months (b) we see longer intervals between births (16 to 24 months).
One can infer a lactation effect, and the short intervals (less than 18 months) may reflect the
earlier return of fertility in women who were not breastfeeding.
114
F 8 . Marital fertility in 1900. Mothers of the birth cohort of 1900 are grouped by age
and cultural origin, to show how many bore another child within 33 months. Among
Protestant women fertility was notably lower and declined with age.
115
.
As we saw in the analysis of mortality, the cultural dimension had a
double impact, and we need to explain both the cultural differences and
their expression at the scale of a neighbourhood. Interpretation is
necessarily more speculative. To crack open the cultural package, we first
propose a representation of infant survival as part of a demographic
system . We then look more closely at the spatial incidence of infant
mortality, and the significance of residential segregation in the politics of
the city. Political discrimination, we argue, affected sanitary conditions,
116
T 12
Estimates for strategic ariables by cultural community, 1900
117
118
119
Differentiated spaces
The urban space was not shared on an equal basis. Nineteenth-century
medical topographers took account of segregation along lines of culture
and wealth as well as the lay of the land, microclimates, and sources of
pollution,%' and Montreal doctors were aware, even at the time of our first
cohort, of the citys excessive mortality , the air poisoning months , and
the deaths of infants and children as a very sensitive test of sanitary
circumstances .%(
In 1900, segregation along linguistic lines was considerable. Most
districts had high percentages of one language group or the other, so that
the pattern departed from a normal bivariate distribution (see Figure 11),
and infant mortality rates in the 50 districts reflected the cultural split. In
1900, since the Anglo-Protestant population so strongly dominated the
high-rent housing stock (over six rooms), and French Canadians the low-
rent stock (see Table 1),%) the overall map of infant mortality also
corresponded in some measure with social class. But we have already seen
that a consistent class effect cannot be detected in the data for individuals.
If we map infant death rates for the two linguistic communities
separately (see Figure 12), we discover anomalies. Some districts show
relatively high mortality for both groups (Hochelaga, parts of Saint-
Henri, and Saint-Jacques). Mortality is relatively low for both groups on
the high-rent flank of Mount Royal, known as The Golden Square Mile ,
120
121
F 1 2 . Infant mortality rates in the urban area. Mortality rates (days 2364) are
separately mapped by 50 districts for francophone sample (above) and anglophone sample
(below), and the histograms show the numbers of districts at various levels of mortality.
122
123
124
A whiff of modernity
The geopolitics which framed the public environment had also an effect on
the attitudes which influenced maternal and family environments. In
reassessing the reasons for the secular decline in both fertility and infant
mortality, scholars have argued the critical importance of attitudes. High
mortality was long fostered by what seemed to be a fatalistic out-
look Its God s will and high fertility likewise as many as God
sends . Reinhard Sprees analysis of inequalities in mortality and
morbidity in Imperial Germany argues for the rise of an instrumental
rationality and an emerging determination to control environments of
home and work.'!
Pertinent to the emergence of such an instrumental rationality are the
mortality experiences of successive generations of parents. Mothers and
fathers of our third Montreal cohort were survivors of a smallpox
epidemic which in 1885 killed 3259 people and left scars, not only on their
faces but on the whole of the body politic. Internationally, smallpox was
the first disease which responded to a public re! gime of vaccination, but
Montreal had not yet put in place the full arsenal of tools required for
125
126
guilty of neglect ; An epidemic of this kind points out some public sin,
some neglect or disorder in the community.'$ But clergy of all faiths
argued that citizens had the duty to use all the means God had given, all
the works of science, medicine and human intelligence, including both
vaccination and cleanliness, to take in hand both the family environment
and the public environment.
Within the next five years Dr Desroches published his CateT chisme
d hygieZ ne : Quest-ce que l hygie' ne ? R : Lhygie' ne est la science qui
enseigne les moyens de conserver la sante! et d e! viter la maladie .'% In the
same interval Dr Se! verin Lachapelle founded his magazine for mothers,
where he, too, urged resistance : Un petit ange dans le ciel !!! While it
might be a word of consolation at a funeral, in a broken-hearted avowal
of impotence, would we, he queried, make such a pronouncement at the
onset of sickness ? Non, mille fois non. Luttons contre le mal .'&
The new attitude, reinforced by advances in science, was reaching an
ever-larger public. Montreal newspapers were reporting in the late 1890s
the decline of mortality from several other childhood diseases , notably
scarlet fever and diphtheria, in France, Britain, and New York City. Their
banner celebrations of the new century honoured Pasteur, and news items
provided week by week instructive examples of cause-and-effect, routes of
contagion, and successful intervention in localized outbreaks of typhoid
and tuberculosis. La Presse, which claimed 68,000 readers, carried full-
page features headlined Les Progre' s de la Science,'' and Dr Desroches
had sold 30,000 copies of his CateT chisme, re-defining contagious diseases
as those dont une hygie' ne bien comprise peut assure! ment nous
pre! server.'(
Expanding initiative can be seen as part of the personal experience of
modernity.') Within the span of birthdates of our third cohort, scholars
have situated Que! bec literary modernity in four nights of poetry readings
and the theatrical soireT es de famille ; 700 young women a month were
writing to the advice columns of La Presse,'* and the French Canadian
business press was pouring forth a rhetoric of progress and initiative,
reaching several thousand retailers, travelling salesmen, and marchands de
noueauteT s (novelty salesmen).(!
Photographs of the infant cohort of 1900 appear, smiling or pouting,
utterly irresistible, in the earliest baby contest of Le Monde illustreT . We
have seen that one-quarter of them died within 12 months of birth, and
that the rate among French Canadians was 42 per cent higher than among
Irish Catholics or Anglo-Protestants. The lines of cultural cleavage had
127
persisted for 40 years, winnowing three cohorts born in the same city. Our
exploration of infant vulnerability, from month to month and year to
year, has shown that poverty is not a satisfactory statistical determinant.
Analyses of seasonality, age structure, and birth intervals point to
systemic properties of cultural practice and to the discriminatory impacts
of residential segregation, ethnic politics, and the political privileges of
property. The French Canadian majority, tightly confined to streets of
high density, was more vulnerable to deficiencies in the public en-
vironment, while the Irish Catholic community, despite limited economic
resources, seems to have achieved some protection for its infants by an
advantageous cultural package and by political and geographic pen-
etration of the turf of the wealthier Anglo-Protestant community.
Despite the persistence of culture-based differentials of mortality,
changes are apparent by 1900, in age at marriage, vaccination, child-
spacing and marital fertility. Newspaper reporting hints at greater
initiative and greater resistance, an instrumental approach to health, and
a whiff of modernity. July 1899 was punctuated by weekly front-page
stories on summer deaths. Over the six months beginning October 1900,
1100 cases of scarlet fever were reported, with 252 deaths, and at the close
of 1900 a count of the years foundlings set their death toll at 97 per cent.("
For the first time, infant mortality was making headlines as a public
scandal, and the recognition that these deaths were preventable would
empower the next generation.
We are grateful for the assistance of the Social Science Research Council of Canada, FCAR-
Que! bec, the Centre interuniversitaire d E! tudes que! be! coises (CIEQ Laval-Trois-Rivie' res),
the Parishes of Notre Dame and Saint Patricks, the Hospital Sisters of Saint-Joseph (Ho# tel-
Dieu), Cimetie' re Notre-Dame-des-Neiges, the Mount Royal Cemetery Company, numerous
archivists, research assistants, and colleagues, in particular, for her help and example,
Danielle Gauvreau of Concordia University.
1 Where in 1880 one-fifth of the sample lived in high-density streets (over 350 households
per kilometre of residential street), in 1900 it reached one-third, including 13 per cent
at densities not observed earlier (5001100 households\km).
2 Matching problems and tests are reported in greater detail in Sherry Olson, Patricia
Thornton, and Quoc Thuy Thach, Dimensions sociales de la mortalite! infantile a'
Montre! al au milieu du XIXe sie' cle , Annales de DeT mographie Historique 1988 (Paris,
1989), 299325 ; Thornton and Olson, Family contexts of fertility and infant survival
in nineteenth-century Montreal , Journal of Family History 16, 4 (1991), 40117 ;
Thornton and Olson, Infant vulnerability in three cultural settings in Montreal 1880 ,
in Alain Bideau et al. eds., Infant and child mortality in the past (Oxford, 1997), 21641.
128
3 Of each set of twins, the second, as shown on the parish register, was excluded from
analyses of next births, and from tables on living conditions of families. Both twins
were included in calculation of death rates, and a dummy variable was introduced into
the categorical analyses.
4 We first combed the census for families with surname B and a child reported under four
years of age. To give adequate representation to the Irish Catholic population, we
added surnames beginning with R for English-speaking Catholics. To find families
where the child had died, we subsequently re-read the census manuscript to check other
married women with surname B and R under 50. Such links cannot always be
confirmed, since the census rarely records maiden names, consequently the census
variables are more often missing for cases of deaths.
5 We combed the records of the cemeteries as well as parish records of burials, and for
Catholic deaths obtained addresses from the original Registres de neT crologie in the
custody of the Parish of Notre Dame.
6 We adapted the method from Jean-Pierre Bardet, Rouen aux XVIIe et XVIIIe sieZ cles : les
mutations d un espace social (Paris, 1983, 2 vols.) The letter B is less sensitive than most
to misspellings, confusion of handwriting, and aural misinterpretation in a bilingual
environment. In a source like the taxroll or directory, unbiased by fertility, it gives
correct weighting to the three principal cultural communities. Although the method
extracts a good sample of the Jewish immigrant population, their infants, as well as
children of Italian and Syrian parentage, are excluded from all analyses, due to
difficulties of census-matching ; in 1901 they amounted in all to 4 per cent of the
population.
7 In cases of mixed marriage we privileged the mothers origins.
8 Because of a general upward shift of rents since 1880 (in near-perfect correlation with
floor area), we modified the thresholds which apply to household rents, street rents, and
occupational medians. Street segments are twinned block-faces. For the cohort of 1900
the low-rent category includes values up to and including $60 per annum, the high-rent
category values of $120 and above. For the 1880 cohort, those threshold values were
both included in the middle-rent category.
9 Data were compiled for 772 street segments (twinned block-faces) in the city and
suburban municipalities (mean size, 85 households), and the segments were aggregated
into 50 districts (mean, 1200 households) by Robert D. Lewis independently of this
project ; he designed the districts to be compact, near-rectangular, and socially
homogeneous. For discussion of the taxroll source and typologies of street segments
and occupations, see David Hanna and Sherry Olson, Social landscape of Montreal
1901 , Plate 30, Historical Atlas of Canada (Toronto, 1990) ; and Sherry Olson,
Occupations and residential spaces in nineteenth-century Montreal , Historical
Methods 22, 3 (1989), 8196.
10 Tests for effects of delayed baptism are reported in Olson, Thornton, and Thach,
Dimensions . In the cohort of 1861, Irish Catholic infants were often baptized at six
or eight days old, French Canadians rarely later than the second day.
11 The first breath is the first sign of life, and a gasp [is] the last.
12 From cemetery records we observed also a trend in the 1870s toward more rigorous
application of the rules for ondoiement.
13 There were in addition 27 stillbirths reported as such ; these are not part of the sample
and do not appear in Table 4, as either births or first-day deaths.
14 Municipal vital statistics become usable about 1900, and distinguish illegitimates,
institutional deaths, stillbirths, and records for the low-mortality Jewish population.
See Annual Report of the Montreal Board of Health (Montreal, 1900).
129
15 Mortality of twins was high (46.5 per cent), as in the earlier cohorts, but the rate of
twinning did not vary significantly among the three communities : just under 1 per cent
as observed over the entire five years. Beyond the first day, we find no effects of mothers
age in the French Canadian sample (where numbers are adequate), but evaluation is
uncertain because linkage of cases to the census (source of mothers age) is biased
toward survivors. By examining survival rates among next-born infants, we observed
that the likelihood of first-day death is increased four-fold (to 20.7 per cent) by a first-
day death of the predecessor, but that bias is removed with removal of first-day deaths.
Where a child survives the first day, the next has a normal chance of surviving it (5 per
cent). The effect may be associated with maternal predisposition to premature birth.
16 Comparative studies have repeatedly shown lower mortality among Jewish infants, and
higher among French Canadian and Polish Catholic infants, than can be explained by
the combination of breastfeeding, family income, crowding, water, and sewer
connections. See U.S. Bureau of Children, Report on infant mortality in Baltimore
(Washington, D.C., 1922) and Robert Morse Woodbury, Infant mortality and its causes
(Baltimore, 1926). For more recent cases emphasizing culture and place, see A. M.
Reid, Locality or class ? Spatial and social differentials in infant and child mortality in
England and Wales, 18951911 , in C. A. Corsini and P. P. Viazzo, The decline of infant
and child mortality (The Hague, 1997) ; E. M. Garrett and A. M. Reid, Thinking of
England and taking care : family building strategies and infant mortality in England and
Wales, 18911911 , International Journal of Population Geography 1 (1995), 69102.
17 Spikes of contagion produced differences in the experience of each cohort. If we look
at the (crude) statistics of the City Health Department over runs of eight years, the 1859
cohort was subject to a normal mortality (summers of 1859 and 1860), the cohort of
1879 to somewhat more favourable conditions (summers of 1879 and 1880), and the
cohort of 1900 to a slightly higher mortality than normal, spread over three summers.
We cannot discern any effect of the year-to-year variation on the variation between
groups.
18 First-day deaths show little seasonality, and were largely attributed in the records to
stillbirths, complications of delivery, or prematurity : 79 per cent of French, 89 per cent
of Irish, and half of Protestant. (The last, from Mount Royal Cemetery, are less
informative records.) The proportions are the same if we examine the entire set of
deaths under one year over the period 1 January 1899 to 31 March 1903. Deaths from
contagious diseases of childhood like measles and scarlet fever show a spring
incidence. For a similar treatment of nineteenth-century data see Jan Sundin, Child
mortality and causes of death in a Swedish City, 17501860 , Historical Methods 29, 3
(Summer 1996), 93106, and for comparable U.S. data in the same cohort (18991900)
see Samuel H. Preston and Michael Haines, Fatal years : child mortality in late
nineteenth century America (Princeton, 1991), 4.
19 Deaths formerly reported in those categories, with strong summer concentration, seem
to have been more correctly diagnosed in 1900 as intestinal. On this problem, see
Gretchen Condran and Rose A. Cheney, Mortality trends in Philadelphia : age and
cause-specific death rates, 18701930 , Demography 19, 1 (1982), 97123.
20 After 1885 the causes displayed always account for between 31 and 39 per cent of all
deaths in the City.
21 R. I. Woods, P. A. Watterson, and J. H. Woodward, The causes of rapid infant
mortality decline in England and Wales, 18611921 , Part II, Population Studies 43,
(1989), 11332 ; Robert Woods and Nicola Shelton, An atlas of Victorian mortality
(Liverpool, 1997).
130
22 Numbers of births in the cohort of 1900 varied little from month to month, but the age-
selective impact of diarrhoea resulted in large numbers of deaths (days 2364) among
those born March through June. We were not able to discern such an effect in 1880.
23 See the suite of papers J. Bourgeois-Pichat, La mesure de la mortalite! infantile ,
Population 6 (1951), 23348 and 45980 ; John Knodel and Hallie J. Kintner, The
impact of breast-feeding patterns on the biometric analysis of infant mortality ,
Demography 14 (1977), 391409 ; and Jan Sundin, Culture, class and infant mortality
during the Swedish mortality transition, 17501850 , Social Science History 19, 1
(1995), 11745. Invariant with overall level of mortality, this linear transformation is
often used as a check on accuracy and completeness of data on infant mortality.
24 Olson, Thornton, and Thach, Dimensions ; Thornton and Olson Infant vulnerability .
25 Populations chosen as reference groups are males, singletons rather than twins, and
French Canadian mothers. Modifying the order of terms in such models has no effect.
26 Valerie Fildes, Breasts, bottles and babies : a history of infant-feeding (Edinburgh, 1986).
For present-day evidence, see Jean Golding, Paule M. Emmett, and Imogen S. Rogers,
Gastroenteritis, diarrhoea and breastfeeding , Early Human Deelopment 49 (1997),
supplement on Breastfeeding : benefits and hazards, pp. S83S103.
27 For an emphasis on local variance of cultural factors in infant mortality see Preston and
Haines, Fatal years ; Hallie J. Kintner, The impact of breastfeeding patterns on
regional differences in infant mortality in Germany, 1910 , The European Journal of
Population 3, 2 (1987) ; 129 ; Alain Bideau, Bertrand Desjardins, and H. P. Brignoli,
Infant and child mortality in the past (Oxford, 1997) ; Sundin, Culture, class and infant
mortality ; Naomi Williams and Chris Galley, Urbanrural differentials in infant
mortality in Victorian England , Population Studies 49 (1995), 40122 ; with respect to
the fertility transition see Michael R. Haines, Social class differentials during fertility
decline : England and Wales revisited , Population Studies 43 (1989), 30523 ; E. Van de
Walle, Communications (Paris, 1986), numeT ro speT cial : De! natalite! , l ante! riorite! franc: aise
18001914.
28 Marvin McInnis, The demographic transition Plate 29, Historical Atlas of Canada,
vol. III (Toronto, 1990). On the adjustment problem, see Danielle Gauvreau, Peter
Gossage, and Lucie Gingras, Measuring fertility with the 1901 Canadian census : a
critical assessment , Historical Methods (forthcoming).
29 Although maternal death rate per 1000 births appears small, the womans risk of
eventual death in childbirth was multiplied by the frequency of childbearing. Following
modern practice, we include as deaths in childbirth cases of reported weakness ,
tuberculosis, and haemorrhage which occurred within three months of a birth.
30 E.-P. Lachapelle et al., Report of the Royal Commission on Tuberculosis. Province de
Que! bec, 19091910. Our figures for causes of death among adults match their
compilation for 18961906, apparently from the same sources.
31 Due to psychological as well as physiological effects, high infant mortality whips
forward high fertility. This is an issue in the literature reviewed by David Reher in Back
to the basics : mortality and fertility interactions during the demographic transition ,
Continuity and Change 14, 1 (1999), 932. For French Canadians in a smaller city, see
Peter Gossage, Families in transition : industry and population in nineteenth-century
Saint-Hyacinthe (Montreal, 1999).
32 We cannot assume natural fertility, although this measure was designed for such a
situation ; see F. Nault, Bertrand Desjardins, and Jacques Le! gare! , Effects of
reproductive behaviour on infant mortality of French Canadians during the seventeenth
and eighteenth centuries , Population Studies 44 (1990), 27385.
131
33 Paul Huck, Shifts in the seasonality of infant deaths in nine English towns during the
19th century : a case for reduced breast feeding ? Explorations in Economic History 34
(1997), 36886.
34 See Valerie Fildes, Infant feeding practices and infant mortality in England,
19001919 , Continuity and Change 13, 2 (1998), 25180.
35 Se! verin Lachapelle, Femme et nurse, ou Ce que la femme doit apprendre en hygieZ ne et en
meT decine (Montreal, 1901).
36 The full 48-month span of observations yields a similar pattern, but a 33-month cut-off
allowed comparison with data for the 1880 cohort.
37 Mapping of Protestant rates reveals fewer infants on the wealthy flank of Mount Royal
than among the immigrants employed by the Grand Trunk railway shops below the
Lachine Canal. The effect may be associated with presence of more older household
heads and their higher rents. Given the higher proportion of next-borns among Irish
Catholic women, the low ratio in a heavily Irish Catholic district seems to have been
limited by factors other than marital fertility ; we suspect later marriage, diminished
immigration, and lower rates of remarriage.
38 As argued by Barry M. Popkin et al. (Breast-feeding and diarrheal morbidity ,
Pediatrics 86, 6 (December 1990), 87482) ; any substitution for breast milk should be
regarded as part of the weaning process. In their extensive research in the Phillippines,
most breastfed infants were given a wide variety of other foods ; and the mere addition
of water and teas doubled or tripled the likelihood of diarrhoeal episodes ;
supplementation with nutritive foods increased the risk as much as thirteen-fold. See
also Dana Raphael and Flora Davis, Only mothers know : patterns of infant feeding in
traditional cultures (Westport, Conn., 1985) ; Rima D. Apple, Mothers and medicine : a
social history of infant feeding, 18901950 (Madison, Wis. ; 1987).
39 J. Villar and J. Rivera, Nutritional supplementation during two consecutive
pregnancies and the interim lactation period : effect on birth weight , Pediatrics 81, 1
(January 1988), 51. See also Y. Mortarjemi, F. Kaferstein, G. Moy, and F. Quevedo,
Contaminated weaning food : a major risk factor for diarrhea and associated
malnutrition , Bulletin of the World Health Organization 71 (1993), 7992. Traditional
ideologies of hot foods , perspiring mothers in harvest season, or sexual relations
which might spoil the milk are widespread (in Ireland and France as well as African and
Islamic cultures), and may lead to earlier weaning. See Fildes, Breasts, bottles and
babies.
40 Kodjo A. Senah, Infant and child mortality : the socio-cultural dimensions of the
problem in Ghana , in Ezekiel Kalipeni and Philip Thiuri eds., Issues and perspecties
on health care in contemporary Sub-Saharan Africa (Lewiston, N. Y. ; 1997), 4990 ; J.
Vanderslice, B. Popkin, and J. Briscoe, Drinking-water quality, sanitation and breast-
feeding : their interactive effects on infant health , Bulletin of the World Health
Organization 72 (1994), 589901.
41 The nursing bottle is the real cause of excess mortality of newborns in summer. The
real remedy, the only remedy, is breastfeeding. Se! verin Lachapelle in La MeZ re et
l Enfant 5 (May 1888), 10.
42 Se! verin Lachapelle, Femme et nurse, 16669 ; in La MeZ re et l Enfant 1 (marsavril
1891) ; and in Le Journal d HygieZ ne populaire, 1 February 1885. Tests in 1899 showed
alarming contamination of the ice supply.
43 See Lachapelle in Le Journal d HygieZ ne populaire, 15 June 1884.
44 Persistence of traditions of infant care is noteworthy. Among Irish Catholic immigrants
to Britain, levels of infant mortality were relatively low despite poverty, and
contemporary British reformers attributed this to higher rates of breastfeeding (see
132
133
Trade in Canada (Report, 189395, vol. 2, Evidence, Quebec, 225) ; on licensed and
unlicensed selling, discriminatory law enforcement, and the means by which property
owners excluded licenses from high-rent streets : Beaver Hall, Richmond Square and
Co# te Saint-Antoine (Westmount).
56 Labor Gazette, 1901. It is difficult to evaluate the local effectiveness of street-cleaning
and sanitary investments. The sewer system became somewhat better connected in the
late 1870s but remained very defective in 1900 ; the number of pit privies was reduced
by half in the decade of the 1890s ; in 1899 there were still 3500 stables in the city not
drained or ventilated .
57 Linteau, Le personnel politique . For a sizable negotiation which affected the Irish
neighbourhood of Saint-Gabriel, see Christopher G. Boone, Language politics and
flood control in nineteenth-century Montreal , Enironmental History 1, 3 (July 1996),
7085.
58 Carpenter, On the vital statistics of Montreal , 154.
59 Removal of household offal was improved in 1899, with some incineration and a sewage
farm ; inspection of milk and ice was gradually tightened. Re-engineering the public
environment was expensive, so that governments devised a much cheaper strategy,
aimed at mothers, as shown by Catherine Rollet in The fight against infant mortality
in the past ; an international comparison , in Bideau et al. eds., Infant and child
mortality, 3860. In Montreal such baby clinics, the Gouttes de Lait, were targeted on
French Canadian parishes, as described by Denyse Baillargeon in Fre! quenter les
Gouttes de Lait, l expe! rience des me' res montre! alaises 19101965 , Reue d histoire de
l AmeT rique francm aise 50, 1 (e! te! 1996), 2968. Childrens hospitals were created primarily
through philanthropy, on segregated lines.
60 Reinhard Spree. Health and social class in Imperial Germany : a social history of
mortality, morbidity and inequality (Oxford, 1988). The importance of attitudinal
change in relation to infant mortality has been demonstrated by Sundin, in Culture,
class and infant mortality.
61 For a balanced account of the epidemic see Michael Farley, Peter Keating, and Othmar
Keel, La vaccination a' Montre! al dans la seconde moitie! du 19e sie' cle : pratiques,
obstacles et re! sistances , in Fournier et al. eds., Sciences et meT decine au QueT bec.
Perspecties sociohistoriques (Quebec, 1987), 87127 ; and for the international
backdrop Peter Sko$ ld, The two faces of smallpox, a disease and its presence in 18th and
19th-century Sweden (Umea University Demographic Data Base Report no. 12 (Umea,
1996). See also Annual Reports of the Montreal Board of Health for 1886 and 1887 ; and
Michael Farley, Othmar Keel, and Camille Limoges, Les commencements de
l administration montre! alaise de la sante! publique (18651885) , 85132 in Peter
Keating and Othmar Keel eds., SanteT et socieT teT au QueT bec XIXe-XXe sieZ cle (Montreal,
1995). An 1884 outbreak in Ontario was contained by a contagionist approach,
interpreted as a decisive shift from miasmatic and mixed approaches to public health by
B. L. Craig, in State medicine in transition : battling smallpox in Ontario, 18821885 ,
Ontario History (1983), 334.
62 Edgar Morin, La Rumeur d OrleT ans (Paris, 1969 ; also published as Rumour in OrleT ans,
trans. Peter Green (London, 1971)) interprets a rumour nourished among adolescent
girls and their mothers, a population excluded from the normal channels of political life.
For analogous political psychology, see William Ryan, Blaming the ictim (New York,
1971). Louis Riel was the leader of a Me! tis rebellion in western Canada. French
Canadians identified with this people of mixed Native and French origins and Catholic
faith. Riel had been a student in Montreal in 1860, and the Oblate missionaries to the
Me! tis were, in 1900, still headquartered at Saint-Pierre-Apo# tre, a parish in the east end
of Montreal.
134
135