increase of more than one person. The rule, therefore, that the average size of the family is not materially affected by including the artificial or arbitrary family does not well apply to towns and cities in which large institutions are located. For the United States the statements given by the census, which includes all arbitrary families, may fairly be taken as representing the average family.
The decrease in the size of families is a subject which causes some alarm. Taking the United States as a whole, it is found by the census figures that in 1850 the average family consisted of 5·55 persons. There has been a gradual decrease, it being in 1860 5·28, in 1870 5·09, in 1880 5·04, and in 1890 4·94. Looking at the different geographical divisions, it is found that this rule holds true except in the Western division, where the average size of the family has risen from 4·18 in 1850 to 4·88 in 1890, the increase having been steady through the intermediate decades. This result would have been expected, of course, on account of the settlement of the "West in the last few years, the population having increased rapidly and being more and more brought to the family basis instead of that of single individuals or young families settling in Western Territories. The small average size of the family in Oklahoma, now a Territory just opened for settlement, shows the influence of new settlements upon the size of the family. In Oklahoma the size of the family will increase until population becomes fairly dense, when it will follow the rule of older communities and decrease. When population becomes more or less urban in character the maximum is reached, and after that a constantly receding average will probably be shown at each succeeding census. A study of one hundred of the principal cities of the country having a population of 25,000 or more, and on the basis of 1880 and 1890, shows with but few exceptions a decrease in the average size of the family. The exceptions are chiefly in the South and West, as might be expected, and as is found regarding those two sections generally. In New York city the average size of the family has decreased from 4·96 in 1880 to 4·84 in 1890, while in Chicago the decrease has been from 5·19 to 4·99 during the same period.
It would be very gratifying if the Federal census statements as to size of families and other social features of population could be carefully verified by independent enumerations. This possibility exists in some cases where States take an independent census. I will call attention to one only, and that the State of Massachusetts, with whose statistics I am more or less familiar. The United States census just taken gives the average size of family in Massachusetts in 1870 as 477. The State census of 1875 gives the average size as 4·60. In 1880 the Federal census shows
