Figures were obtained from some other colleges, which are incomplete and should be taken with reservation. Their incompleteness probably led the number of children to be considerably underestimated. At Amherst, 1872-1879, it was found that 44 of the 440 graduates of the period remained unmarried. The average number of children per married man was 1.72. At Wesleyan it was found that 20 of the 208 graduates, from 1863 to 1870, remained single; the average number of children per married man was 2.31. The only satisfactory study of the birth-rate of graduates of men's colleges is that recently made by John C. Phillips from the class lists of Harvard and Yale, 1850-1890, summarized in the accompanying graph (Fig. 37). In discussing his findings, Dr. Phillips writes: "Roughly, the number of children born per capita per married graduate has fallen from about 3.25 in the first decade to 2.50 in the last decade. The per cent of graduates marrying has remained about the same for forty years, and is a trifle higher for Yale; but the low figure, 68% for the first decade of Harvard, is probably due to faulty records, and must not be taken as significant. "The next most interesting figure is the 'Children Surviving per Capita per Graduate.' This has fallen from over 2.50 to about 1.9. The per cent of childless marriages increased very markedly during the first two decades and held nearly level for the last two decades. For the last decade at Yale it has even dropped slightly, an encouraging sign. It is worthy of note that the number of children born to Yale graduates is almost constantly a trifle higher than that for Harvard, while the number of childless marriages is slightly less." This is probably owing to the larger proportion of Harvard students living in a large city. If the birth-rate of graduates both of separate men's colleges and of separate women's colleges is alarmingly low, that of graduates of coeducational institutions is not always satisfactory, either. To some extent the low birth-rate is a characteristic of educated people, without regard to the precise nature of their education. In a study of the graduates of Syracuse University, one of the oldest coeducational colleges of the eastern United States, H. J. Banker found[123] that the number of children declined with each decade. Thus married women graduates prior to the Civil War had 2 surviving children each; in the last decade of the nineteenth century they had only one. For married men graduates, the number of surviving children had fallen in the same length of time from 2.62 to 1.38. When all graduates, married or not, are counted in the decade 1892-1901, it is found that the men of Syracuse have contributed to the next generation one surviving child each, the women only half a child apiece.