REPORT ON THE CRINOIDEA. 51 



formula — a. 3. (3). This method, a determination of the characters present in the majority 

 of cases, is the only one which can be safely relied on for fixing the characters of a species ; 

 and it is therefore apparent that the formulae given by both Bell and myself for species of 

 which we have only seen single individuals are necessarily liable to subsequent correction. 



Bell has encountered this difficulty of irregular arm-divisions, and has met it by 

 giving three formulae for one species which he names Actinometra variabilis. 1 It seems 

 to me that two would have been sufficient, as the characters indicated by the first, 

 A'.3.2., are also expressed in the third, A'.3.(2).(2) ; while there must be a considerable 

 mistake somewhere ; for Bell's first and second formulas do not provide for more than 

 forty arms, though he gives the total number of arms as sixty to ninety. His second 

 formula is A'.3.3., which of course represents a very different type from A'.3.(2).(2). 

 So far as one may judge from his figured specimen, the last is much the most correct, 

 for out of thirteen palmar series only two consist of three joints. On some part of every 

 ray there are three divisions above the palmare, each, with but one exception, consisting 

 of two simple joints. I find that a similar arrangement presents itself upon each of the 

 other three specimens of this type, and I should therefore write its formula as — a.3.2.2.2.2, 

 not using brackets for the last figure because a fifth post-radial axillary occurs in each of 

 the four individuals examined. Neither of Bell's formulae, however, allow for more than 

 three post-radial axillaries, while his second one A'. 3.3. would indicate by the absence of 

 brackets a type with exactly forty arms, and regular distichal and palmar series of three 

 joints each all round the cup, i.e., such a form as Actinometra parvicirra, while in 

 reality Actinometra variabilis only resembles that species in the constant presence of 

 three distichals, its later arm-divisions being totally different from those of that type. 



While therefore it is extremely desirable to be able to examine a good number of 

 individuals before attempting to describe and give a formula for any new specific type 

 of multibrachiate Comatulae, I do not think that there is any serious objection to describing 

 a species from one individual only. For so far as the characters of the arm-divisions are 

 concerned, I have found it to be an almost invariable rule that the characters which 

 present themselves most frequently in any one individual are those which distinguish the 

 species. Thus, for example, bidistichate series only presented themselves in five out of 

 twelve specimens of Actinometra parvicirra, 2 in which the number of distichals is 

 typically three. Two of these individuals were certainly abnormal, the numbers of 

 bidistichate and tridistichate series being exactly equal. But in the other three specimens 

 the largest number of bidistichate series was three, and they never presented themselves at 

 all in seven individuals. The same may be said, though with a somewhat less degree of 

 certainty, respecting the palmar series, sixty-seven of the seventy-six present consisting 

 of three joints. Palmare only occurred in eight of the twelve specimens examined, and 

 were abnormal in but four of them, one species being unusual in having three two-jointed 



1 "Alert" Report, p. 155. 2 Trans. Linn. Soc. Lond. (ZooL), 1879, ser. 2, vol. ii. p. 44. 



