Sources of Information . 259 
reached inductively as to the normal number that drop out at 
each grade. 
There are, therefore, four groups of tables in the study of the 
two cities, with a fifth group which compares the conclusions 
thus reached with the facts collected from other cities and 
writers. Admitting errors in each method employed, second¬ 
ary tables and considerations have been introduced: on the one 
hand to eliminate errors so far as possible; on the other hand 
to determine the limits of probable error, so that if per cents 
could not be determined with absolute accuracy, statements 
could at least be made as to the maximum and minimum limits 
within which the truth lay. 
Neither through my present acquaintance with statistical 
methods nor through the co-operation of a university instruc¬ 
tor in statistics, have I been enabled to find methods ready¬ 
made which were applicable to this problem. The only resource 
was one not uncommon in scientific work, a slow process of 
trial and experimentation, during which more time was spent 
upon the rejected methods than upon those finally adopted. 1 
SOURCES OP INFORMATION. 
To assist any who may wish to use the same material, and 
also to explain the data upon which my inferences are based, a 
brief statement may be made at this point as to the sources of 
information. A large portion of the data concerning the Chi¬ 
cago schools was taken from the Annual Reports of the Board 
of Education, beginning with the year after the Great Fire, 
1871-72, which destroyed all the previous records. Since the 
year 1888-89, unfortunately, the school enrollment by grades 
has not been printed, — which fact necessitated (in 1894 at the 
1 Thanks are due to Dr. Hourwich of the University of Chicago for 
valued assistance; also to Mrs. Stevens, statistician of the Chicago Board 
of Education, in allowing me to copy the annual returns in advance of her 
own elaboration of them for the Annual Report, and especially in loaning 
me the bulky files of the years 1889-93, which are the only data now in 
existence from which may be obtained the grade enrollments for these 
four important years. Most important of all was the co-operation of Mrs. 
Folkmar on the Milwaukee portion of the work, since she not only elabor¬ 
ated the most of the material but added important theoretical suggestions. 
