sub-zone increased from 50.0 to 86.4 per cent French, while the agricultural sub-zone went from 79.6 to 92.2 per cent French.
The St. Maurice valley, almost entirely French from the beginning, dropped very slightly, but the counties of the Laurentians adjacent to Quebec City increased from 93.8 to 97.7 per cent. Metropolitan Quebec increased from 75.2 per cent French to 94.3 per cent, while Montreal hardly changed from its 64.6 per cent of 1871. The Saguenay and Lower St. Lawrence show very slight change, while the Gaspé area shows an increase from 64.9 to 82.7 per cent, a similar change to Saguenay County on the other side of the river.
Particularly striking is the increase in the counties of Quebec near Ottawa where the percentage French rose from 45.8 in 1871 to 77.3. This is similar to the change in the Ontario county of Russel from 50.4 to 81.9 in the same period.
Table 2 shows, for each zone of the province, the percentage French to the total population for each census from 1871 to 1951. There is a striking uniformity in the changes which have taken place over the 80-year period. For example, the industrial portion of the Eastern Townships represents the largest increase, and its per cent French, in every decade without exception, gains between 2.5 and 7 per cent. Oscillations are only to be found in the table where the net change has been very slight. The Montreal Plains industrial counties show an increase from 92.1 to 94.3 per cent over the 80 years and this includes four decades of increase and four of decrease.
Trends as uniform as these would seem to offer opportunities for prediction safer than those presented by most demographic data. It seems clear that if industrialization continues at a high rate, the number of French in urban areas will increase as a result of migration. Whether or not the present rate of industrial growth continues, the differential birth rate demonstrated by other students 2 will have the effect of increasing the proportion French in both industrial and farm areas of the province.
Sociologists see the spatial distribution of groups such as the French and English in Quebec as the unplanned result of individual movements in which people take up the location to which they
PERCENTAGE FRENCH TO TOTAL POPULATION FOR ZONES OF THE PROVINCE OF QUEBEC, CENSUS YEARS 1871-1951
are in some sense best suited. An ecological process known as succession has resulted, as large French families bought farms from the English whose families were smaller, for example in the Eastern Townships 3. According to one of Horace Miner's informants, about one-quarter of the farmers of St. Denis place two sons on farms in each generation 4.
For a study of succession in agricultural areas one should ideally have data on individual farms in each locality at each census : whether they are operated by English-speaking or by French-speaking farmers, their family sizes, and hired help, and most particularly, the departures of individual family members. Lacking these, we use the changing residence of the population as a whole  to derive estimates of actual movement. The breakdown of the movement into French and English must await further study.
2. AN ESTMATE OF THE MOVEMENT
FROM AGRICULTURE TO INDUSTRY
Retour à la table des matières
The large families of French Canada are seen in many ways. They are the hope of the future, the sign of a robust and vigorous national life, the result of moral principles and the assurance of its continuance. These are not the concern of a purely demographic paper, which takes up numerical aspects only. Many writers have alluded to the fact that if farms are not to be divided, then there must be a single inheritor of the family lands in each generation. The non-inheriting children can remain on the farm as dependents or can leave to found new farms elsewhere, or they can go into non-farm occupations. In Quebec, if they leave the family farm and become farmers elsewhere, they must either buy land, in general from the English, or settle new territories. If they leave agriculture they may become priests or storekeepers, or go into city factory work. The various logical possibilities are shown on Chart 1.
The possibilities of this scheme are rather well known. Not scholars alone but every person and every family deals more or less consciously with the arithmetical paradox arising when lands are inherited by one child and families are large. Any circumstance which affects individual families so directly is sure to-stand out in popular consciousness and to be a favorite subject of discussion and of literature. In this respect, the migration from the land is quite the opposite of the differential in family size (discussed further in Part 4 of this paper), the latter being visible only as a statistical difference between groups whose family sizes on the ground do not look noticeably different.
Although the phenomenon of migration from farms is well known, it has not been measured. No census volume or statistical yearbook shows the number of French-Canadian young men and women leaving their parents' farms. To have objective measurement would be valuable, for the actual amount of movement may turn out to be much greater or much less than estimated by popular and scholarly guesses. In forecasting the amount of industry that
will be needed to continue to take up the flow from the farms, we require a knowledge of what the flow has been in the past in relation to past birth rates.
The importance of measurement, however, does not by itself make measurement possible. Statistical results sufficiently accurate to be useful, for the past period with which we are concerned, may simply not exist, – it is possible for such a phenomenon to be lost without trace. This is fortunately not quite the case. The movement from farms with the growth of a factory economy underlies those facts on the number of persons residing in different parts of the province at successive censuses which we have discussed in Part 1, and by the use of these facts of residence and other data we can infer how many people must have moved. To do so involves assumptions at several points, which we hope to cover with suitable safeguards. It may be well to describe the general strategy before we deal with specific figures.
The essence of the strategy is the use of two more or less independent approaches. Admitting the arbitrariness of the assumptions in any one method, we use two methods and consider that the difference between them is at least a first step in measuring the error of each.
The first method is confined entirely to 13 counties of the province. In all of these, the population is largely engaged in farming. They are the main source of migration to cities, and, what is essential to our calculation, their areas are either fixed through the period from 1871 to 1951 or their populations can be adjusted to the equivalent of fixed areas. Initially we work from the numbers of persons at the several ages found in the counties, census by census, and make assumptions in regard to mortality among them. If, in the ages 15-59 there were 64,000 men in 1901, and if new entrants (from the age group 5-15) less deaths equal 18,000, and the 1911 population is 68,000, then 14,000 men must have left. Net migration is thus a residual figure in the reconciliation of successive censuses. It is expanded by the ratio of all farms in the province to those in the 13 counties, to constitute our Estimate 1 of the province-wide migration from farms.
The second approach is through the number of persons working in agriculture and in non-agricultural industry. The fact that non-agricultural industry rose by 748,000 while agriculture dropped  17,000 between 1891 and 1951 should provide a clue to the movement. The change over the whole period, however, conceals a difference that arises very conspicuously in the 1940's. During the war and post-war years, the population in agriculture in the Province of Quebec dropped from 252,000 to 188,000, a decline of 64,000. This decline more than counterbalanced the steady rise that had been shown from 1901, and hence the surprising result that, although the Province of Quebec is almost three times as great in population in 1951 as it was in 1901, it contains fewer men in agriculture. The increase in non-agricultural industry is shown in every one of the thirteen main occupational groups, except fishing and trapping which, like farming, declined sharply. The rise from 79,000 to 237,000 in manufacturing occupations is especially conspicuous (Table 3).
We first estimate the natural increase and the net movement for the province as a whole, and then calculate the proportion of this natural increase which arises on farms. By subtracting from the natural increase on farms the actual increase of farm workers between successive censuses, we obtain our Estimate 2 of the net outflow from agriculture.
If methods 1 and 2 produce similar figures, this is evidence that both are realistic. We will not bother the reader here by giving the arithmetic of their computation but will remain satisfied with presenting Table 4 which shows the results of the two methods 5.
In total and in five of the eight decades, agreement is surprisingly close. The main discrepancies are shown in 1871-81 and in the decades containing the first and second world wars.
To what extent does the degree of agreement between the two methods check either of them ? Agreement does not demonstrate the suitability of the death rates used, since the same rates entered both calculations. Their appropriateness is however to be judged by the fact that they produce a number of deaths lower in the total of all ages than those officially recorded for the Province of Quebec, rural and urban together. Differentials between the youngest ages of life and older ages, and between rural and urban parts, may be such that the survival rates here used are not too high for the favoured group of rural men of working ages. Any
MALES GAINFULLY OCCUPIED, 10 YEARS OF AGE AND OVER,