members of groups ancestrally widely divergent should have converged and assumed general characters so similar as to be scarcely distinguishable from each other.

On page 356 Cambridge writes of the family Psechridae that "in all probability they constitute a cribellate form" of the Agelenidae. On the other hand, on page .'ib'O he writes of the family Uloboridac:

The spiders comprehended in this family probably have no near relations amongst existing forms; so that it is difficult to place them in any system of classification. They represent possibly the survivors of larger groups, the majority of whose members have long since died out.

And yet Koch, Thorell and McCook did not hesitate to consider the genera l loborus and Hypliotes as belonging to the family Epeiridae.

Dahl does not attribute great value to the eribellum and calamistrum. We have seen that he relegates the Hypochilidae together with the Avieulariidae and Atypidae to the sub-order Tetrapneumones. This alone would show that he considers the possession of a eribellum and calamistrum as a subordinate character and the structures themselves as of later and independent origin. We find further evidence of this attitude in the separation of the various cribellate families and their inclusion under different sub-orders. The Xoropsidae which Dahl once treated as a single family lie now has split into three families. Of these the Zoropsidae, containing, according to Dahl, a single genus Zoropsis, he places in the sub-order Citigradae, the Tengellidae in the sub-order Polytrichiae and the Zorocratidae in the sub-order Tubitelae. Moreover. Dahl puts the Oeeobiidae as a sub-family under the family Urocteidae and the Acanthoctenidae as a sub-family under the family Ctenidac. If a taxonomic system is not an arbitrary separation of forms on the basis of characters which have no value whatsoever, but represents an attempt at expression of true relationship, and if we analyze this classification of Da Ill's, we must come to the conclusion that the separation into his suborders anteceded the formation of the eribellum and calamistrum and that therefore Dahl is in favor of the second theory mentioned by ('ambridge.

Have the ancestors of Arachnoniorph spiders possessed two pairs of luminous pleiopods, as the recent Liphistiids still possess them? Have some of them then developed a eribellum from the endopodites of the anterior pair of pleiopods, while others either entirely lost these endopodites or changed them into a eolulus? Have these two types then developed independently with convergence in other characters due to functional adaptations.? Or is the eribellum the original structure and have the Ecribellatae gradually developed from various families of the Cribellatae? Or is, as Cambridge states, the cribellum a new structure independently developed in different families?

The last two questions must be carefully kept apart, for the answer to them would lead us to very different conceptions in classification. Suffice it to mention that in the one case the Oecobiidae would have developed from the Urocteidae and the Acanthoctenidae from the Ctenidae, while in the other case the Urocteidae would have developed from the Oecobiidae and the Ctenidae from the Acanthoctenidae.

Paleontology unfortunately gives us no clue whatsoever to the solution of all these questions. Arachnomorph spiders seem to have existed already in the Carboniferous, but the number of species and, in fact, of specimens which are known to us is so small and the specimens themselves so imperfectly preserved that we do not know whether those spiders had a cribellum and calamistrum or were ecribellate, whether the number of their spinnerets was already six or still eight, as in the Liphistioniorphae. We know spiders again only from the Tertiary. The American specimens are too imperfectly preserved to give us a fair idea of the structure of the spinnerets and hind-legs in every case. But the spiders from the Amber of the Baltic and some species from Switzerland and France are so well preserved that one may discern all the most important structures. We can state, therefore, positively that in the Tertiary both Cribellatae and Ecribellatae were already represented and that quite a number of families were already well defined.

We have, therefore, to seek an answer elsewhere, and first of all from the evidence furnished by embryology. Although a great deal remains still unknown and should be investigated, and other things are not convincing and should be reinvestigated, yet on the whole we have a fairly good picture of developmental changes. It is fairly safe to assume that structures which are common to several families and which appear early in the development represent ancestral characters antedating the separation of the ancestral stock into the present families. At the same time it is quite possible that some character having more general value, and therefore greater stability than other characters, undergoes modification at a later period. But it is more than doubtful whether so-called atavistic reversals to ancestral characters can ever take place if such characters have once been permanently lost. In examining developmental changes of spiders we find that the cribellum develops from the endopodites of the abdominal pleiopods of the fourth abdominal somite. They are really the pair of spinnerets still in existence in Liphislius, modified, but functioning in all cribellated spiders. In forms with a divided cribellum the change has not yet progressed so far as in forms with an entire cribellum. The latter is a later stage in evolution. In such genera as Dictyna and Acanthocephalus we still find species with a divided cribellum and such with an entire cribellum. The colulus is only a further step in the disappearance of the endopodites accompanied by a loss of spinning tubes, the last step being the suppression of the colulus and the complete approximation of the spinnerets.

Embryologically the cribellum appears comparatively early and is present in spiderlings before they leave the cocoon. Similarly the colulus is developed before even the traces of abdominal segmentation have disappeared. To give an idea of the changes which take place in a spider between the completion of its embryological development in the egg and maturity, I shall give here a brief description of a spiderling of Lycom hell no taken at random from a cocoon and let the reader compare it with a specimen of a mature spider of the same species or at least with a description of it as given by Chamberlin in his excellent monograph of Xorth American Lycosidae.

At this stage the spiderling has neither eyes, nor hair, nor spines. Its body and appendages are smooth and colorless. The shape of the carapace is approximately the same as that of the mature spider, though I have not subjected it to careful measurements. The sternum is oval, but is transversely w^ider than long. The fang of the chelicerae is the only chitinized structure. It is extraordinarily small, as wide at base as long, and about as long as one-half the diameter of the basal article of the chelicerae. The maxillary plates are clearly defined, but do not have the shape as in mature spiders. The colulus and the spinnerets are very well developed. The anterior spinnerets sit somewhat in advance of the other spinnerets. The anal tubercle is quite prominent. It is oval and above it the dorsal intersomitie lines are still in evidence. Both lung-slits are in their proper positions and the rudiments of lungs already discernible. The tracheal spiracle is clearly visible in front of the spinnerets and the bases of the two tracheal steins are also discernible. There are as yet no traces of the genital rima. The segments of the legs are clearly defined, but do not have the final proportion. The trochanteres are still without a notch. There are three claws on each tarsus, but the claws are quite smooth.

If we now take a spiderling of the same species from the mother's back we find already some important changes. There is as yet no pubescence of any kind, but pigmentation has appeared on the carapace. The eyes are present, but their arrangement is that of a Pisaurid and not of a Lycosid. The sternum is now narrower, but is still wider than in the adult. Intersomitie lines can still be discerned above the anal tubercle. Pubescence appears after the first molt. The margins of the chelicerae and the spines on legs assume, however, their final structure only after three more molts.

Taking all the evidence into consideration, we may assume, therefore, that the ancestors of Arachnomorph spiders all possessed a cribellum and had three claws on their tarsi. The latter is evidenced by the fact that spiderlings of two-clawed species possess in their early stages three claws, and that a rudimentary claw is still present on the anterior tarsi of some Zoropsids. The calamistrum, on the other hand, must have been a somewhat later acquisition and, once acquired, persisted until the present <lay. It disappears only in the males of various families after the last molt, being clearly and perfectly defined in all earlier stages. It also becomes more irregular in the two-clawed Zoropsidae and in the Tengellidae—a condition evidently subsequent to its development and leading to its final disappearance.

The similarity in structure and habits of Cribellate and Ecribellate spiders may be readily understood from this point of view. It is not convergence, nor is it spontaneous development of specialized yet identical structures in phylogenetically different families. // is the development of ecribellated families from cribellated ones through the fusion and transformation of the cribellum into a colulus. Rather than imagine the separation of cribellated spiders from ecribellated ones and the subsequent independent development of the two divisions, ivc have before us a com plicated process by which at different times both new cribellated and ecribellated families were produced from the original three-clawed cribellated ones. It is safer not to unite cribellated and ecribellated forms under a single family, as Dahl did it in the case of the Urocteidae, even though it seems quite reasonable to consider the latter a younger branch of the Oecobiidae. One may be asked, why, then, not segregate the cribellated families under a single division of Cribellatae, as Simon does it? But such a segregation would imply the separation of the Arachnomorph spiders at an early date into the two divisions, whereas my contention is that no such separation took place at one single time. Moreover, the process of evolution must have embraced ecribellated families which in their turn gave rise to new ecribellated families.

The presence of a calamistrum in spiders with a cribellum does not present any great difficulties. It represents nothing but modified hair and varies in length and arrangement. Such spiders as some Zoropsids with a more or less irregular calamistrum give us a clue to its origin. In the cribellated ancestral stock from which Tertiary and Recent Arachnomorph spiders developed, the cribellum may have had simply the shape of an indefinite stretch of somewhat modified hair. Gradually the irregularity gave place to a single or double row of specialized hair, while in forms which lost the cribellum, the ealamistrum. instead of developing, completely disappeared.

The organs of respiration present a greater difficulty. We know that the tracheal system has developed from the book-lungs and is of a more recent date. Besides the Hypochilidae with two pairs of lungs, we have the Dysderidae and Oonopidae with one pair of lungs and one pair of tracheal spiracles; we have the Caponiidae and Telemidae without lungs, but with two pairs of tracheal spiracles and we have the bulk of spiders with one pair of lungs and a single tracheal spiracle. The single median spiracle undoubtedly developed from a fusion of a pair of spiracles, first through the formation of a transverse common groove uniting the two spiracles and serving as a sort of a shallow atrium, and then through an approximation of the spiracles, together with a shortening and deepening of the uniting groove. The first step in this direction may still be observed in the Filistatidae. The somitic position of the spiracle remains always the same. Its distance from the lungs is of secondary importance and is entirely dependent upon the growth of the third abdominal somite. The position of the spiracle both in the Argyronetidae and Anyphaenidae considerably in advance of the spinnerets must therefore be regarded as the older one.

The transformation of the first pair of lungs into tracheal tubes must be much more recent. We may imagine such a transformation taking place without considerable modification of other characters. Nevertheless, I cannot agree with Fage when he waves aside this difficulty and unites the Tejemidae in one family with the Leptonetidae. while at the same time separating from them the Ochyroceratidae. The latter have scarcely any important structural difference from the Leptonetidae and should form a sub-family of these, while the Telemidae undoubtedly deserve the rank of a family.

The position of the eyes and their number are of considerable interest. Since it has been established that both may vary within the same family and even genus, their value as a character has been considerably diminished. Moreover, the arrangement of the eyes has been studied entirely by systematicians and not by morphologists. The direct eyes, usually recognizable from the others without microscopic dissection, have been designated as the anterior median eyes. In the majority of cases these eyes, together with a pair of indirect eyes, form a slightly recurved or procurved line which is spoken of as the first row of eyes, while the other

« PreviousContinue »