One extensive statistical inquiry, however, has been conducted by Dr. J. M. Rice.1 Conspicuous not only for the singularity of its presence within the field of pedagogical discussion, but equally so for the skill and discernment with which it was carried to a conclusion, this investigation has done much to clear up vague opinions as to the place of spelling in the elementary school, and to establish many important facts concerning the effect of the age, environment, etc., of the pupil, and of the methods and other factors of teaching upon the results of instruction. The writer has found this investigation very suggestive, and has employed some of Dr. Rice's tests in experiments to be described later.

The phenomena of spelling offer material for psychological investigation and analysis, apart from any pedagogical application. Spelling, whether oral or written, depends upon the association of certain arbitrary symbols obtained through one or more sense channels, the subsequent recall of these symbols in a conventional order and the expression of this ordered group of symbols by some form of motor activity. Spelling is a sensori-motor habit which expresses itself in every concrete instance of the spelling of a word as a synthetized motorial reaction following, at more or less remote temporal intervals, certain complicated sensory stimulations. Errors in orthography are more instructive than the correctly spelled words. Granting that a child has at some time seen, or heard, or otherwise received sensory impressions of the constituent elements of a word in their proper order, a misspelling of such word may be viewed as a mild form of aphasia, using that term in its broadest signification, as any disturbance, however slight or temporary, of the function of language. This question will be treated more in detail in the consideration of the various classes of spelling errors.

Enough has been said in the foregoing pages to emphasize sufficiently: (a) the great importance ascribed to spelling and the consequently large expenditure of time, money, and energy 1 "The Futility of the Spelling Grind," Forum, April and June, 1897.

devoted to this branch of school instruction; (b) the lack of definite and reliable information concerning the factors and results of this training, and the necessity, therefore, from the pedagogical point of view, of careful inductive inquiry that the most economic disposition of the educational forces may be made to secure the largest results; (c) the independent psychological interest that may attach to experimental and statistical research in this field.

PART I

THE VARIATION OF THE NORMAL OF ACCURACY FOR AGE, SEX, SCHOOL GRADE, AND GENERAL MENTAL CAPACITY

THE material upon which the present statistical study is based was collected from pupils of several Philadelphia schools and embraces the following kinds of tests:

1. Lists of spontaneously selected words - the maximum number that could be written in fifteen minutes.

2. Lists of specially selected words written in vertical columns from the dictation of the teacher.

3. Series of short sentences (containing selected words) written from the dictation of the teacher.

4. Spelling papers of the regular term examination set by the superintendent of schools.

5. Compositions and other written exercises of the regular schoolroom work.

To obtain the lists of spontaneously selected words, the pupils were given conveniently prepared slips of paper and told to begin at a given signal and write as many words as they could, using any words at all that "they might happen to think of." The words were written in vertical columns, and at the end of fifteen minutes the signal to stop writing was given and the papers were collected. The papers were written by the pupils of the third- to the eighth-year classes, inclusive. Several such series of papers were taken, one at the Foxchase School, by the principal, Dr. Oscar Gerson, at the end of the school year, June, 1896; and at the Northwest School a series was taken at the end of the school term for three successive years, viz.: June, 1896, June, 1897, and June, 1898.

Though these papers were not secured originally for a spelling investigation, they lend themselves readily to this purpose. Moreover, owing to the conditions of the test, they afford some specially interesting data, and have, therefore, a peculiar value as compared with the usual spelling statistics. The latter are generally based upon lists of words selected by the examiner; by the conditions of the test here described, the pupils were compelled to select the words from their individual vocabularies.

The object of the experiment having been to give as much freedom as possible to the flow of ideas, the pupils were informed that the test was in no sense a test of scholarship, and were directed to write as fast as they could. An examination of the papers has shown that these directions produced the desired result, viz.: a list of words freely chosen and representative of the dominant idea groups in the subjects' minds. But when it was decided to examine the orthography of these papers, it was thought that the conditions under which they were obtained were so unfavorable to legible writing and correct spelling that the percentage of mistakes would be very high. The results have shown this very natural assumption to have been wrong. Though some of the papers show the effects of an attempt to write as rapidly as possible, legibility as well as sense (the latter measured by the general character of words selected and the greater or less use of mere nonsense words or syllables) being completely subordinated to the evident purpose of writing a great number of words, yet, on the whole, the percentage of orthographical errors, illegible and nonsense words is very small. In all the series taken the percentage of correctly spelled words ranged from 89% in the third-year classes to over 99% in those of the eighth year, while the illegible or nonsense words are but a fraction of one per cent of the total number written..

Great care was had to secure uniformity of conditions in taking the several series of fifteen-minute lists. Some preliminary experimentation had shown that extreme precautions

THE NORMAL OF ACCURACY

were necessary if the results obtained were to be regarded as at all comparable inter se. A slight difference in the wording of the directions given to the subjects of the experiment, or even in the manner of the experimenter, produces notable differences in the character of the series written. To limit as far as possible the influence of this personal variation, a simple set of explicit directions, as short as was consistent with clearness and as free as possible from the influence of suggestion, was prepared and was read to those participating in the test. That the directions met these requirements is evidenced by the fact that out of more than fifteen hundred papers, less than a dozen were rejected because of the pupils' evident misapprehension of the conditions of the test.

The several series from the Northwest School were taken each year by myself, as it seemed inadvisable to have the class teachers make the tests, even though they would have been limited to the reading of the printed directions. Dr. Gerson conducted the collection of material in the Foxchase School. Further uniformity was secured by taking the several series at about the same time of year, towards the close of the school term in June, and by using the same kind of materials for making the lists - lead pencils and specially prepared sheets of paper. Although the majority of the subjects of the first test participated in those of each successive year, yet they had been advanced each year to the next higher grade. The personnel of each class of a large school changes also more or less during the year by the withdrawal of some pupils and the entrance of others. In comparing, then, the results of the three successive tests of the Northwest School with one another, these facts are to be borne in mind. That is, variations in the average results must be attributed, in some part at least, to the variations in the composition of the classes furnishing the word lists.

The collated results of the several series are exhibited in Tables I to IX (Appendix, pp. 71-75). In Tables III, V, and VII the boys' and girls' papers have been treated separately