2. Number of anal rays. 



In Report I, 1913, p. 5, mention is made of the fact that the number of anal 

 rays is very large and variable; that they may be very difficult to count with 

 accuracy, and -that these purely practical difficulties had induced me hitherto to 

 dispense for the most part with the fin-rays, though I did not doubt that the 

 number of rays might characterize the species. 



These difficulties have now been partly overcome by the employment of well 

 preserved specimens of elvers stained with alizarine. By this means, the rays, or 

 at any rate their basal parts, in the unpaired fins are rendered so distinct as to 

 permit of their number being determined with absolute certainty, though it will 

 naturally always be a lengthy and laborious task to procure checked countings of 

 elements occurring in such great numbers. In addition to this, the difficulties may 

 be greatly increased if the staining, for some reason or another, should not be 

 completely successful. The prepared specimens of European elvers were quite 

 satisfactory from a ttechnical point of view, whereas in the only two samples 

 available of American and Japanese elvers, the staining of the anal rays was for 

 the most part but poor, which rendered the work of counting far more difficult. 



The material examined consisted of 1.365 European, 245 American and 73 

 Japanese elvers. The result of the countings is shown in Table III. From this it 

 will be seen, that the series of figures obtained for the three species overlap. 

 Closer examination shows that Angiiilla rostrata has the lowest values, Angiiilla 

 japonica the highest, the majority of variants in the case of the first falling below, 

 in that of the last above 215', whereas Anguilla vulgaris occupies an intermediate 

 position, with approximately equal numbers of variants above and below 215. 

 Similarly, we find that the lowest average is that of A. rostrata, (198.649), the highest 

 that for A. japonica (220.26), whereas the average figure for A. vulgaris, (215.224) is 

 .intermediary, lying nearest, however, to that of A. japonica. 



The table shows, in addition to the average for each sample, also a (standard 

 deviation), P. E. A. (probable error of average), and P. F. A. (probable fluctuation 

 of average). This last is, as in Report I, 1914, calculated by multiplying the pro- 

 bable error by 5, and thus indicates with a probabihty of 1300 — 1400 to 1 the 

 limits within which the actual average of an unlimited number of specimens of 

 the single groups would lie. 



We see from the table, that A. rostrata differs decidedly both from A. 

 vulgaris and A. japonica, its P. F. A. lying quite outside that of the two others. 

 On the other hand, A. vulgaris and A. japonica resemble each other more. Never- 

 theless, the average in the case of A. japonica (220.26) is far outside the P. F. A. of 

 A. vulgaris (214.223—216.225). The wide range of the P. F. A. of A. japonica 

 (216.13—224.39), owing to which the P. F. A. of the two species are not entirely 

 distinct is doubtless merely due to the paucity of the Japanese sample as compared 

 with the European (73 as against 1.365). Even a comparatively slight increase in 



' It must be regarded as due to mere chance, that no variants below 200 have been found in 

 the case of A. japonica; the material however, consisting of onlj' 73 specimens, is very small in com- 

 parison with that for the other two species examined, especially A. imlgaris. 



