528 TRANSACTIONS OF SECTION D. 
11. A rudiment of the right oviduct is present. 
Sepiadarium agrees with Ididsepius in numbers 1, 4, 6, 7, and 9; in regard te 
the other points its position is :— 
2. The connection of the mantle and head resembles that of Sepiola; a rudi- 
mentary cartilage is present; the mantle is fused with the base of the funnel on 
either side. 
3. The fins are sub-circular, situated half-way back along the mantle. 
6. The pallial nerve is unbranched, and traverses the stellate ganglion. 
8. There are small anterior salivary glands, and the posterior are fused as in 
Rossia. 
10. The left ventral arm is hectocotylised at the tip, and the spermatophores 
are attached to the buccal membrane. 
11. There is a single left oviduct. 
Septolotdea is very inadequately known, but it agrees very closely with Sepia- 
darium in those parts which have been investigated. 
There can be no question that the above assemblage of characters allies these 
three genera to the Sepiolidé rather than to the Sepiide or Loliginide ; in fact 
practically the only character of any importance which points in the opposite 
direction is the hectocotylisation. This portion of the subject has been very fully 
and clearly handled by Dr. Appelléf of Bergen (1), and to his memoir I refer 
those who desire more detailed information. We have here, then, a case in which 
forms whose ventral arms are hectocotylised are more nearly related to forms 
with dorsal hectocotylisation than to others with ventral, and this shows that the 
position of the modified arm (or arms) is not by itself an infallible guide to 
systematic affinity. It is a striking instance of an aphorism of the late Professor 
Rolleston, that ‘no single character can be regarded as a safe basis for a natural 
classification until it has been proved to be so.’ 
It may, however, be worth while to look a little further into the relationships 
of these forms, and to see whether the hectocotylisation of the dorsal arms is quite 
as sporadic and irregular as it at firstappears. It must be remembered that in by 
far the larger number of the Decapoda the hectocotylisation affects the ventral 
pair of arms. If we take the possible lines of development of this group, we must, 
I think, admit that the Gigopsida are more primitive than the Myopsida; the 
paired oviducts furnish the most obvious proof of this. The closure of the ocular 
aperture would seem to have been the first step in the evolution of the Myopsida, 
for this character is present in the group without any-exception. This primitive 
Myopsid stem divided into two branches (2). With one, leading to Loligo and Sepia, 
we need not concern ourselves further. The other gave rise to the group of forms 
we are at present considering. These seem to have had a short stumpy body and 
sub-circular fins; the right oviduct was gradually becoming abortive, and the shell 
was undergoing atrophy; the mantle was still connected with the back of the 
neck by a cartilaginous joint; a complete muscular liver capsule was being 
formed ; the posterior salivary glands were fused, and the hectocotylisation affected 
the ventral pair of arms. The next step in differentiation was the separation of a 
branch leading to Idiosepius which was given off before the right oviduct had 
entirely disappeared, and still showed a primitive character in the posterior 
situation of the fins. It must, however, have acquired independently the fusion 
of the mantle with the head in the nuchal region, and the separation of the 
posterior salivary glands; for though these reappear in other members of the 
family, they are not found in Rossza. 
As to the further evolution of this group two possibilities present themselves :— 
A. The main stem divided into two branches leading to Rossia and Sepiola 
on the one hand, and to Sepadarium and Sepioloidea on the other. 
B. The stem gave off first a branch leading to Rossia, and subsequently divided 
into two, one leading to Sepiola and the other to Sepiadarium and Sepioloidea. 
