In the case of the serological identification of yeasts, the direct application of the serum reaction, which can be used in bacteriological differentiation, seems to be unreliable. Hence the interrelationship between the several yeasts was examined by the writers by means of the reactions of agglutination, precipitin and complement fixation. The yeasts used were as follows: Endomyces fibuliger Lindner, Zygosaccharomyces Barkeri Saccardo-Sydow, Debaryomyces globosus Klecker, Saccharomyces sake Yabe (sake yeast), bottom beer yeast "Asahi," Schizosaccharomyces octosporus Beijerinek and Torula C. Yeasts which are included in different genera according to the botanical classification, were clearly distinguished from each other by the agglutinin test. Especially, a high agglutination reaction was recognised in both Debaryomyces globosus and Torula C. In the cases of Saccharomyces species as sake yeast and beer yeast, however, their aggultinating powers were very closely related. Employing Fornet's "Ringprobe" to carry on the precipitin test, the writers have used the maceration juices prepared from the yeasts modifying Lebedew's method. The maceration juice was prepared as follows: The yeast cells which had been grown in Koji extract were filtered, washed with water, and dehydrated by spreading them on clay plates. The dried yeast powders were mixed with water in the proportion of 0.5 mg : 1 cc., then the mixture was incubated for 24 hours at a temperature of 30℃. The clear juice separated by centrifugation was diluted with equal volumes of 1.7% salt solution. From the examination with the maceration juices, it was shown that while Debaryomyces globosus and Torula C reacted specifically, other yeasts could not be distinguished from each other, since they accompany groupreaction in this test. Therefore, it seems that in general it is not possible to distinguish the interrelationship between the yeasts merely by the precipitin reaction e mploying "Ringprobe." Furthermore, the writers have employed "Absattigungsverfahren," modified by Misao, to differentiate the yeasts for w hich the "Ringprobe" method returns unimportant results. This convenient method has given comparatively positive results for the differentiation of the yeasts stud ied. But the differentiation of sake yeast and beer yeast was not so clear was as the case with other yeasts; therefore, the writers conclude that the two yeasts are very closely related to each other. It makes the technique easier and gives more precise results to employ "Absattigungsverfahren" than the complement fixation. From the serological point of view, it was shown that (1) sake yeast and beer yeast are in closest relationship; (2) Zygosaccharomyces Barkeri is also fairly closely related to Debaryomyces globosus ; (3) Endo myces fibuliger reacts positively in almost the same degree against Zygosaccharomyces Barkeri, Debaryomyces globosus, sake yeast and beer yeast respectively, and vice versa; (4) Torula C reacts positively against Zygosacch aromyces Barkeri, Debaryomyces globosus, sake yeast and beer yeast, but negatively against Endomyces fibuliger; (6) Schizosaccharomyces octosporus is also negative as against all the yeasts.