_Fourth._--The direct examination of the child bears specially upon his state of instruction and degree of intelligence. The inspector comes to a positive decision with regard to each child, and asks the opinion of the doctor, as well as of the head of the special school, who assists. _Fifth._--The inspector has a schedule of particulars regarding the children finally accepted for the special school filled up by their teachers in the schools from which they came. The medical examination will be considered in the next chapter. Let us add, in conclusion, that all the decisions arrived at are to be regarded as provisional; the children are to be admitted to the class for defectives on trial, to be kept under observation. FOOTNOTES: [6] See _Année Psychologique_, vol. xii., p. 1, and vol. x., p. 116. The method sometimes adopted, for other purposes, of asking the teacher to classify the children according to their intelligence is quite fallacious. Teachers make no allowance for age. Recently an excellent teacher pointed out to us, as the most intelligent in the class, a child who had really, when his age was taken into account, a retardation of two years; but in a class of younger children his age gave him an appearance of mental superiority. [Such facts vitiate much statistical work on the correlation of "brightness" in school-children with other qualities.--TR.] [7] Teachers have a troublesome habit of saying simply "attendance regular" or "irregular." The inspector should demand an exact return of the absences. [8] There are two methods of stating the representative value of a group, the _average_ and the _median_. Everyone knows the average. The median is obtained by arranging the values in linear series from the smallest to the greatest and taking the middle one. When should one use the average, and when the median? It is not easy to give a general rule, but in this case of spelling, we have a good example. If we wish to calculate the number of mistakes for each age, to take the average might be a disastrous proceeding. A single child who made a hundred or so mistakes would obviously make the average unfairly high. The median is affected much less by such aberrant cases, and consequently is more suitable for very heterogeneous series, in which the difference between the maximum and the minimum is very great. [9] By way of comparison, the following dictation was given to ninety-two children in an Edinburgh school. The progressive difficulties depend upon the non-phonetic spelling and the lesser familiarity of words. Most of the children came to school in their sixth year. 1. Tom is a good boy. He has a book and a bat. He can run fast. 2. The dog is bigger than the cat, but he cannot climb so well. He would if he could. 3. The farmer walked through the wood till he came to the field. It was a fine day for sowing the corn. He hoped it would not rain till he had finished his work.