former is preferable to the latter as the measure in our present study. Inasmuch, however, as the arithmetical mean is the one in most common use, and further, as some may still feel that it, if investigated, would show the rise that has been supposed to exist, we will consider the data and results that Table III. shows. In this table are shown
|U. of Ala.||21-0.0||20-1.8||20-2.4||20-3.6||20-6.0||949|
|N. Y. U.||21-1.6||21-2.3||20-8.4||21-7.5||21-10.8||860|
|Av. of Totals||23-1.3||23-3.4||23-0.8||23-2.3||23-1.9|
the arithmetical averages of each college by decades, supposing that the students graduating at any given year of age, say 32, are about equally distributed throughout the months of the year, thus giving an average for the given year of, say 22.5 years. With small numbers,
this assumption is not without its liability to error; but with numbers so large as we have, the errors are found by actual trial practically to negative each other; so that we can rely upon the results as being, for all practical purposes, and in the main, substantially correct.