In the same year, Caldwell and Courtis, having adapted and distributed some of the questions from the Boston School Survey of 1845 to school districts around the United States in 1919, could claim that education in 1923 was superior to that of three ever quarters of a century earlier. As Travers (1983) observed, however “the data do not bear close examination” (p. 156). The groups of children were not really comparable, and neither were the tests and the scoring procedures. The 40,000 children who took the test in 1919 did deter on some of the test questions, notably in geography and history, and worse on others, notably in arithmetic, with the median score on score on all questions being somewhat higher in 1919 than in 1845. Caldwell and Courtis had chosen the five easiest arithmetic questions to repeat and had even supplied some additional information (“A rood is a measure of land on longer in use. There are 40 sq. rods in a rood and roods in an acre.”), but, for example, the question “what part of 100 acres is 63 acres, 2 roods, and 7 sq. rods? Was answered correctly by only 16% of the sample in 1919, whereas 92% of the Boston students surveyed in 1845 had answered it correctly without having to be reminded what a rood was. Part of Caldwell and Courtis’s argument was that such test questions dealing as they did with striations unlikely to arise in the children’s experience, had been eliminated from modern textbooks and were therefore of little value for comparison purposes. This issue-the adequacy of comparisons based on test questions that might be more appropriate for one group of students than another-would arise again some decades later when national and international surveys of mathematic achievement were launched.