cancel two specified characters with maximal speed and accuracy. The time required for each task was recorded in seconds, and each individual was scored in terms of the number of items correctly checked.

Sixty-four Chinese students were individually tested. They were divided into four groups of approximately equal numbers on the basis of the order in which the tests were taken.

The average values per individual for the four groups and for each of the various tests are given in Tables I. and II. In

the first table the values represent the number of seconds required for each task; hence the smaller values indicate the greater speed. The values in the second table represent the number of items correctly reported or checked; and in this case the larger values represent the greater accuracy or efficiency. The values for the vertical and horizontal arrangements for each test are given in adjacent columns to facilitate ease of comparison.

These group results are convincing and unambiguous. Each of the four groups read and cancelled Chinese characters with greater speed and accuracy when they were arranged in vertical columns, while better results were invariably achieved when the English letters and Arabic numerals were arranged in horizontal sequence.

The majority of the individuals in each group conformed to the group-preference in each of the eight comparisons with but one exception; viz., Group I in the test for accuracy in cancelling numbers. Eight of these sixteen individuals gave better records for the horizontal arrangement; six did better with the vertical arrangement, and two did equally well in both.

Similar results are obtained when the four groups are regarded as a unit. Ninety per cent. of the individuals read, and eighty-nine per cent. cancelled, Chinese characters faster with the vertical arrangement, while ninety-five per cent. cancelled English letters and ninety-four per cent. cancelled Arabic numerals faster with the horizontal arrangement. The corresponding percentages for accuracy are sixty-nine, sixty-two, sixty-one, and sixty respectively.

The majority of the individuals in each group conformed to the group norm in all four of the tests for speed, or regarding the groups as a unit-seventy-five per cent. of the individuals conformed to the group-averages in all four tests for speed. The individuals are less consistent with respect to accuracy; for here only twenty-eight per cent. of the total number of individuals conformed to the average group-results in all four tests.

There is no convincing evidence that the preference of these individuals for the vertical arrangement of Chinese characters is stronger than that for the horizontal arrangement of English materials. While the gross differences between the group-averages for the two arrangements of Chinese characters are generally larger than those for the English materials, the comparative results are quite similar when stated in relative or percentage terms. Moreover, there is but little difference in the number of individuals who consistently exhibit the two types of preference.

These preferences are more pronounced for speed than for accuracy. The differences between the group-averages for the two arrangements with respect to speed are larger, both absolutely and relatively, than those for accuracy. On the basis of speed, a considerably larger percentage of the individuals manifest the two preferences than is the case when their performances are measured in terms of accuracy (see above).

These factual results justify the conclusion that the majority of these Chinese students apprehend Chinese characters more quickly and accurately when they are vertically arranged, while they apprehend English materials more readily and correctly when they are arranged in horizontal order. The two preferred arrangements, however, are much more conducive to speed than to accuracy of apprehension.

As has been indicated, there are numerous exceptions to these two preferences. Some of these may be due to chance factors, some to differences of difficulty of the two materials, and others to the order of presentation. A few of these exceptions are so striking that they are suggestive of individual differences. For example, one individual invariably made better records with the horizontal arrangement of the Chinese characters.

TABLE III

RELATION BEtween LengTH OF TRAINING IN ENGLISH AND THE NUMBER OF EXCEPTIONS TO THE GROUP NORM

The suggestion that many of these exceptional results are due to individual differences is supported by the fact that the number of exceptions per individual is correlated to some extent with length of residence in America and the number of years in which English has been studied. Table III. gives the

number of individuals who deviated from the group norm in a specified number of the eight tests, together with the corresponding average number of years' residence in America and the average length of time in which English has been studied. Fourteen subjects deviated from the group norm in three of the eight tests, and these individuals on the average have studied English for 10.57 years and have lived in America for 4.85 years. Forty-four subjects gave exceptional results in less than three of the eight tests, and these individuals have studied English and have lived in America for a considerably less time than the twenty individuals who deviated from the group norm in three or more of the tests.

Undoubtedly, the number of deviations from the group norm is directly correlated with length of residence and amount of training in English.

TABLE IV

RELATION OF LENGTH OF TRAINING IN ENGLISH AND ABILITY TO READ CHINESE IN VERTICAL COLUMNS

For

The corresponding data for the deviations from the group preference for the vertical arrangement of Chinese characters and for the horizontal arrangement of English materials are listed separately in Tables IV. and V. respectively. These tables state the number of exceptions in the four tests. example, twelve individuals deviated from the normal preference for the vertical arrangement of Chinese characters in two of the four tests, and thirty-two subjects did not conform to the group-preference for the horizontal arrangement of English materials in but one of the four tests.

There is a direct correlation, with one exception, between the number of deviations from the group-preference for the vertical arrangement of Chinese characters and length of

residence and number of years devoted to the study of English. Moreover, the degree of correlation is apparently the most pronounced for those with the greater amounts of training in English. In other words, a considerable amount of training in English enables these students to apprehend Chinese characters as readily with one arrangement as with the other.

TABLE V

RELATION OF LENGTH OF TRAINING IN ENGLISH AND ABILITY TO READ ENGLISH WITH A HORIZONTAL ARRANGEMENT

There is also a direct but small correlation between the number of deviations from the normal preference for a horizontal arrangement of English materials and the length of training in English. The differences between the averages are too small to be of much significance. Taken at their face value, however, these results lead to the somewhat surprising conclusion that the relative ability of these Chinese students to apprehend English materials in the horizontal direction actually decreases to some extent with practice.

It has frequently been suggested that the preference of the Chinese for the vertical arrangement of their characters is ultimately due to some constitutional condition of the eyes that enables them to apprehend materials more easily and readily when they are arranged in this fashion. It is at once evident that the influence of such a hypothetical condition can readily be overcome by training, inasmuch as these students can read English with a greater facility when it is printed in horizontal columns.

It may be suggested that the best arrangement is a function of the materials; that the Chinese characters are so constructed that they can be more readily apprehended when they are arranged in vertical columns. Such an hypothetical influ