Having thus determined upon these five classes as representative and significant, one hundred college students were asked to ponder these classes in connection with the subjoined array of the common uses of the word “good,” to choose one of the classes as the one most appropriate in each case, and then to choose as many other of these classes as seemed to be involved, and to rank them in order of their importance. For example, they were shown the first two terms of the array,—(1) “Good food; fit to eat, untainted,” and (2) “Good food; nutritious, palatable,” and were asked whether they belonged in Class A, B, C, D, or E. It was at once seen that membership in more than one class was implied in both cases, but it was also admitted that nutritious and palatable food (No. 2) was a more dependable “good” than was the food specified by example No. 1. First choices were consequently indicated on this basis. After that, second, third, and succeeding choices were made for Nos. 1 and 2; following which, the remainder of the array was treated in the same manner. The average of the results derived from this experiment indicates that Class B (dependable “good”) has a majority of votes, it being given first place 49 times, and a subsidiary place 75 times. Class C was given first place 20 times, and a secondary [Illustration: _CHART_ _Showing the distribution of the 79 significations of the word “Good.”_ (_Explanation in text._) _Key_ = _First Choice._ = _Second Choice._ = _Third Choice._ = _Fourth Choice._ = _Not Chosen._ _Classes A B C D E_] place 65 times. Class A was the first to be implicated in 5 uses of the word “good,” and was mentioned as a later choice 26 other times. Class D was voted to be the one most obviously implied only 2 times, but was mentioned in an associative relationship 59 times, while Class E, which received only one vote, got that for first place. The importance of Class B is thus clearly apparent. However, an almost equally significant fact seems to be that there were only five cases in which one class alone was implicated, whereas two classes were used in 18 cases, three classes were used in 34 cases, and four classes were simultaneously implicated by 20 terms of this array. The appended chart renders this distribution and overlapping more obvious. From all this, moreover, we seem to be warranted in inferring that, with all the unlikeness there is between these various synonyms for “good” when taken one at a time, when taken all together (_sub specie aeternitatis_, as Spinoza would say) there is more of a common core of meaning in them than one might naïvely expect. Consequently, we may safely presume that if we can satisfactorily define any one of the preceding five classes of “good” in physiological terms, we shall at the same time have hinted the definitions of all the remaining classes.