SAGITTA GAZELLAE 249 



illustration of this point, but again this only differentiates S. lyra "lyra "-type from S. lyra " gazellae" '- 

 type and S. gazellae. 



The posterior fin of S. lyra is often shown with a forward 'insertion' into the anterior fin (e.g. 

 Ritter-Zahony, 191 1, fig. 3) and this appearance is often seen; it is due to part of one pair of lateral 

 nerve cords crossing the lateral fields between the fins (Fig. 7), and I have been unable to detect a 

 similar feature in S. gazellae ; it is not, however, a particularly obvious feature in most specimens of 

 S. lyra. 



Tail segment. Another feature of difference which has been much discussed is the relative length of 

 the tail segment. Ritter-Zahony's (191 1) tables show an overlap of this feature in the two species, but 

 this overlap may be due mainly to uneven shrinkage (see p. 240), and the method of compiling tables 

 giving tail percentages and head armature formulae gives undue weight to exceptional specimens. For 

 example, if in a series of ten specimens of similar length nine give tail percentages between 1 1 and 

 12% and one gives a measurement of 15% then the table will record the tail percentage for this size 

 group as 11-15%, m no wa Y taking into account the abnormality of the one specimen in comparison 

 with the other nine. (This becomes even more apparent when dealing with hook numbers and, in 

 Fig. 8, I have tried to avoid this difficulty by using a scatter diagram which shows all the observations 

 including the abnormal ones, thus enabling the reader to assess the value of each one.) 



The interpretation of the situation in regard to tail segment percentages given by Michael (1919) 

 and quoted by Burfield (1930) is a fair statement: 'the tail segment percentage is usually less than 

 15% in S. gazellae, and usually more than 15% in S. lyra\ 



It is evident from Table 7 and the preceding discussion that there are in fact recognizable differences 

 between S. lyra and S. gazellae. The characters upon which those differences are based, however, 

 cannot always be determined with sufficient precision to justify the retention of the two as separate 

 species. However re-examination of S. lyra and S. gazellae has resulted in some further characters, 

 some of which are reliable ; these are summarized in Table 8 and are discussed below. 



Table 8. Diagnostic characters of Sagitta lyra and S. gazellae from re-examination of the species 



DIAGNOSTIC CHARACTERS FROM RE-EXAMINATION OF THE SPECIES 

 Total length. S. gazellae reaches a maximum size of 105 mm. in Antarctic waters, and 68 mm. in 

 Subantarctic waters ; the largest specimen of S. lyra among 1 593 specimens from the ' Discovery ' 

 collections was 42 mm. long. Previous records of S. lyra at a greater size than this are possibly due 

 to confusion with other species ; for example, Michael (191 1) did not recognize S. maxima as a separate 

 species (see p. 252), nor did Germain & Joubin (1916) report S. maxima in the Monaco collections, 



