estimates of intelligence and of the quality of schoolwork were made and compared with the results of the psychological tests themselves. Of the latter there were two batteries of so-called general intelligence tests: the first, the army alpha, is a verbal test based quite extensively upon language and dealing with concepts of verbal nature; the other, a slightly modified army beta, adapted to school children who understand at least a modicum of English necessary for instructions. Rather than Rather than selecting the school children from certain grades the writer selected all the twelve-year-old children in the schools which avoided the biased sampling that must have arisen had children in one or two grades only been selected. By this method, then, all the twelve-year-olds from the entire public school systems in the communities surveyed were studied irrespective of their grade location. Incidentally the manner in which these twelve-year-olds distributed themselves through the grades indicates once again the fact that chronological age is no criterion to the pedagogical age of a child, of whatever racial stock.

This is no place for a detailed report of the findings in this study, but the pertinent facts for the problem of immigrant mentality will be briefly given. It was found that the actual grade location of the children is the best single criterion of their intelligence aside from the mental tests themselves. The comparative standings of the Latin with the non-Latin groups shows that the modal grade for the latter is the low seventh, for the South Italians, the high fifth, for the Spanish-Mexicans, the high fourth, for the Portuguese, the fifth. On the basis of the teachers' estimates of general intelligence on a seven-rank scale the Italians are on the average 8 of one class-rank below the "Americans," the Portuguese are over one class-rank below the latter, and the SpanishMexicans over one and one quarter class-ranks below.

14

Coming directly to the results of the tests themselves, the important fact for us is that of overlapping of intelligence in the groups. The following tables give first the absolute differences as measured by the tests, then the measurement of the overlapping :

14 The Portuguese group was not segregatable into half grades.

(B) ALPHA RESULTS

SHOWING PERCENTAGES OF FOLLOWING RACIAL GROUPS EXCEEDING THE QUARTILES
OF THE AMERICAN GROUP (Actual and theoretical) 15

Per cent. exceed.
American Q1

Per cent. exceed. Per cent. exceed.
American Median American Q3

Race group

(Actual) (Theor.) (Actual) (Theor.) (Actual) (Theor.)

From the average for the American twelve-year-olds it looks at once as though measured by the alpha the Latins are decidedly inferior to the Americans. The beta averages, while more nearly alike for the Latins and Americans, still show a marked superiority for the latter. It must be noted that the range of the beta units is just half that of the alpha, hence the differences in the actual figures in the table are not indicative of the real differences revealed. It is still true, however, that the Latins approach their competitors more closely in the beta than in the alpha tests. Thener latter is based, as we noted, upon verbal concepts, while the latter is a so-called performance test operated independently of language. Were it not for the differences shown in other measures of intelligence, as, for instance, grade location, teachers' estimates of intelligence, and school marks, it might be imagined that the differences brought out in beta are more truly significant of the differences in ability than alpha. I have shown elsewhere that this is not the case,18 but space prevents a lengthy discussion of the matter here. The following points summarize these facts:

First, the beta tests proved too easy for the American twelveyear-olds who were above their years in intelligence. This means that the tests were not diagnostic enough for these children and that there was a piling up of high scores for these children at the upper end of the scale. This narrowing of the spread of their actual abilities is the reason for the lower P.E. of the Americans in beta as compared with the Latins. Had the tests been more suited to their real mental capacities, some of them would have scored higher and the average for their group would have risen, and their variability as a group would have been greater. For the Latins the tests proved well suited for their capacities.

15 The figures tabulated under "theoretical" are the computed measures of overlapping when the reliability of the tests are known. Cf. Kelley, T. L., "The Measurement of Overlapping." Jr. Educ. Psychol., 10: 458ff.

16 Discussed in writer's study mentioned in footnote No. 13 above.

Second, checking the alpha by correlation against the outside criteria, such as, grade location, teachers' estimates of intelligence, and the school marks, it was found that the alpha, the verbal test, was more diagnostic of the mentality and hence educability of the Latins and the Americans both than was the beta. In addition these correlations go to indicate that the asserted language handicap under which the foreign children are supposed to labor does not exist, at least so extensively as imagined." As Miss Thomson put it the failure in the tests and in school is not based so much on failure to understand as failure to comprehend.

The measures of overlapping (sections B and C of above table) reveal more significant facts of mental differences in these groups than the differences in average score. If these tests are at all true measures of differences, and there is certainly much evidence that they constitute the best measure of mentality science has yet produced, may we not assume that the extent of the overlapping among the groups clearly concerns an important problem before us, namely: the relation of these differences to future racial mixture which is subsequent to our enormous immigration? If there is anything preponderant in mental heredity it appears that this question must be answered.

Taking the South Italian group as a sample of the Latins and near Latins (as we may term the Spanish-Mexicans),-if they equalled the American group, 75 per cent. of them should exceed the first quartile or 25 per cent. of the latter. As a matter of fact in alpha less than 25 per cent. of them exceed the lowest 25 per cent. of the Americans. So, too, with the median or second quartile, but 7 per cent. of the South Italians exceed the lower 50 per cent. of the Americans, and but 1 per cent. exceed the upper 25 per cent. of the latter.

It may be that alpha is unjust to the Latins even though there is considerable evidence that it is on the whole quite adequate for diagnosis of mental differences. It may even be that the language difficulty which a proportion of the Latins certainly had when they entered the public school has persisted until its effects are

17 It is rather interesting to recall that thirteen years ago Ayres in his book, Laggards in Our Schools, pointed out that language difficulty did not operate as a factor in school retardation. He wrote: "Wherever studies have been made of the progress of children through the grades, it has been found that ignorance of the English language does not constitute a serious handicap." (p. 116). Of course no one would argue that in children newly arrived in this country from foreign countries speaking another language would not be handicapped in the period of acquiring our tongue. But beyond that period, the trouble in retardation lies in other directions.

felt as late as five or six years after entry."

Therefore for those

who would still maintain that alpha is too distinctly advantageous to American children, the overlapping in beta, in spite of the fact that the evidence is clear that this test is unfair to the American twelve-year-olds, still shows the latter superior. But 33 per cent. of the Italians exceed the lowest 25 per cent. of the Americans, but 18 per cent. exceed the lower 50 per cent. and but slightly over 8 per cent. the upper 25 per cent. There is no evidence that the beta was in any way unfair to the Latins either in content or presentation. All the children were tested by the writer personally in small groups and in each session members of all the racial stocks were present. The Portuguese and the Spanish-Mexicans compare less favorably with the Americans than the South Italians.

We may now conveniently summarize the results of the experimental findings on this subject before passing to interpretative paragraphs.

The army tests on recruits and draftees showed that the Latins and other Southern European (also certain Eastern European) groups were decidedly inferior to the men who were natives of Northern and Northwestern Europe. The Dickson-Thomson studies have presented good evidence that the intelligence of the racial groups, at least of those investigated, does not improve by education, and further that in the samples which they study the Latin groups are on the average eight tenths (.8) the average intelligence of the American children of North European ancestry. Murdoch's survey revealed that as measured by the Pressey scale of general intelligence, the Italian groups which she studied were decidedly inferior to the American and Hebrew children tested, and even slightly lower than the negro children measured at the same time. Her twelve-year-old group (which we noted was a selected' sample) compares quite noticeably with the writer's. In her results she reports that 13.5 per cent. of the Italians equaled or exceeded the median of the Americans. The writer's figure for a similar measurement of overlapping was but 7 per cent., but the difference may be due partly to the nature of the test and partly to the fact that his sample was not at all handpicked. It may be due to actual differences in the intelligence of the group. The important point, however, is the likeness of the two studies.

18 The writer in answering those who maintain that all differences in mentality (barring the extreme feeble-minded) are due to environment and opportunity, is always puzzled to know how these persons account for the distribution of children at 12 years of age, some in third, fourth, fifth, sixth and seventh grades, when they all started in school at the same time.

One may use norms which were developed1o for translating the alpha scores of school children into Binet mental ages. If this is done it is found that the Latins measured are on the average two years retarded as compared with the American groups studied. Again the caution must be repeated that the range of scores in each group is wide, meaning that the differences are relative not absolute.

Space forbids further comment on the studies themselves, but we must turn before concluding to a few paragraphs on the probable bearing of these findings upon the problem of immigration and the mixture of immigrant stocks in this country.

If the researches into mental heredity have the validity which we suspect, then the differences shown in these investigations just cited are simply reflections of what is found in the general adult population. The findings of racial differences in immigrant stocks revealed in the army lend weight to this assumption. Further it is interesting that the preliminary testing of the Japanese and Chinese made in California indicates that these Oriental immigrant stocks compare very favorably with the American population of North European ancestry in the same neighborhoods. Surely the language handicap is of greater potency in the Oriental than in the European. So it seems that these methods of testing have enough pertinence to make us pause and consider the meaning of their findings.

As noted at the outset of this paper, we have shifted the source of our immigration from North to South and East Europe, and further we have drawn more and more upon a low type of peasantry as compared with the rather good average industrial and agricultural classes that formerly came to us from German, Scandinavian and Anglo-Saxon countries.

As we noted above20 the problem of mixing racial stocks is one of producing high average intelligence in the masses so as to lay the biological chances for the production of each generation of the superior persons who will make the greatest advances for their generation. If the mentality of the South Europeans who are flooding this country is typified by the mentality of the three groups studied by the writer and others, the problem of future standard of living, high grade citizenship and cultural progress is serious. The writer does not attempt to say whether these findings are

19 From norms developed by Drs. S. C. Kohs and W. M. Proctor at Stanford University (1918-19). Cf. Proctor, W. M., "Psychological Tests and Guidance of High School Pupils." Jr. Educ. Research, Monog. 1, No. 1, pp. 70 (1921).

20 supra, p. 5.