REPORT ON THE TETRACTINELLIDA. 
evil 
single character, but on an assemblage of characters ; in this respect the genera of Tetillidse 
are very excellent genera, but in several of the other families they are not so ; thus I have 
separated the genus Myriastra from Anthastra among the Stellettidae on account of 
the presence of two forms of aster in the latter, the former possessing only one, but this 
may possibly be justified by the great importance of the microsclere as a character in 
classification. The existence of a difference in the number of distinct kinds of spicules 
present in different species or the replacement of one distinct form of spicule by another, 
as an aster by a microrabd, frequently enables us in the absence of associated differences 
to group species together into assemblages, which are of great convenience, and which 
might perhaps be termed, as they have been by previous describers, subgenera ; certainly 
if the presence or absence of spines over the surface of a spicule is a character of specific 
value, the absence of a distinct form of aster, or the substitution of a toxa or a micro- 
strongyle for an aster, is of something more than specific value ; whether we call it generic 
or subgeneric does not signify. Passing from genera to families it may be expected that 
here at all events we shall stipulate for the absence of intermediate forms as a condition 
to their separation ; to this again I respond “No.” So long as we admit the persistence 
of genera and species through long intervals of time, so long does it appear to me 
must we admit the possibility of transitional forms occurring between groups of all orders 
of rank, though, it need hardly be said, that the higher the rank of the group the less 
likely are such transitions to occur, and the more primitive the group, other things being 
equal, the less likely are they to occur, for the simple reason that divergence between 
the groups of higher rank and primitive forms took place at a much more ancient date 
than that between groups of lower rank and higher forms. 
But we shall justly expect to find families defined rather by an assemblage of 
characters than by any single one, though naturally some single character will be more 
useful than others, and we find in the Sponges, as in so many other instances, that this 
is not usually the most physiologically important character, but rather the least so ; 
owing possibly its preservation to this very fact, which has put it beyond, or almost 
beyond, the scope of selective influences. 
Having formed our families on a consideration of the sum total of their differences 
and resemblances, we shall be in a position to judge of what are the most constant 
characters, and a knowledge of these will help us in the formation of higher groups, and 
in the investigation of the family relationships of other Sponges which are not so well 
known as the Tetractinellids. While the characters found to be constant in one group 
may thus serve as guides to the classification of another, they are merely guides, and so 
treacherous that they must be “followed with a loaded pistol at their heads” ; in other 
words, a character, which may be constant enough in one family, may be a notorious 
example of variability in another. We shall now attempt to discover what are the most 
constant characters in the families of the Choristida. 
