North Africdu Speviinenfi o/ A[»iis cniicrifoiniis. 501 



scattered ii'rc<j,\ilarly over the wliole or })art of the dorsal 

 sui'face. (The ventral surface is always smooth.) 



(b) 111 tAvo there is an additional lar<i,e median dorsal 

 spine (fig. .2_, D). 



(6") Several have lateral groups of denticles. 



{d) The spines on the outer angles may be from one to 

 foui". Usually there are two large spines, but there may be 

 one large one overlapping a smaller one as in A. inauritanicus, 

 thi-ee equally large, or two large and two small. 



The Scottish specimens also are variable in respect to the 

 marginal spines, though none have the scattered surface- 

 denticles noticed above. The marginal spines are arranged 

 differently in almost every individual, and I have seen also 

 in some cases an additional large median dorsal spine 

 (fig. 2, B). 



I conclude from these facts that the differences in respect 

 of these characters relied upon for the separation of Ghigi's 

 species are really within the limits of normal variation of 

 A. cancriformis. 



There are, however, other characters which are regarded 

 as specific, and may be touched upon : — 



(1) Length of the longest flagellum of the first leg. 

 Typically this flagellum reaches about to the end of the dorsal 

 shield. In A. apulius it reaches to the base of the telson, and 

 even exceeds the whole body in A. mauriianicus. Now all 

 my Algerian specimens except one agree, in that the flagellum 

 reaches but little beyond the shield ; but in this one, and 

 in all those from Scotland, it readies to the base of the 

 telson. It is evident that this character is not so constant 

 as has been supposed. 



(2) Number of legless segments. This is undoubtedly 

 tlie most reliable specific character for use in this difficult 

 genus, but even here there is a limited variability. According 

 to Simon and others, there are, in A. cancriformis, 6 leg- 

 less segments in the female and 7 in the male. But I find 

 among twelve of my Scottish specimens that only three have 

 6 and nine have 7 of these segments. Among the Tunisian 

 and Algerian individuals the majority of females have 7 and 

 the males have 8. One of the characters given for A. mauri- 

 ta/ncus is the possession of 7 legless segments in the female 

 and 8 in the male, and it is clear that this is not in itself 

 of much impoitance. Braehm (1893) examined a large 

 material, and found that these legless segments vary from 

 5 to 7 in the female and from 6 to 8 in the male. 



(3) The number of segments uncovered by the dorsal 

 shield is of tin given as a spcciiic character,, though it is 



