TRILOBITES THE MOST PRIMITIVE ARTHROPODS.

The Arthropoda, to make the simplest possible definition, are invertebrate animals with segmented body and appendages. The most primitive arthropod would appear to be one composed of exactly similar segments bearing exactly similar appendages, the segments of the appendages themselves all similar to one another. It is highly improbable that this most primitive arthropod imaginable will ever be found, but after a survey of the whole phylum, it appears that the simpler trilobites approximate it most closely.

That the trilobites are primitive is evidenced by the facts that they have been placed at the bottom of the Crustacea by all authors and claimed as the ancestors of that group by some; that Lankester derived the Arachnida from them; and that Handlirsch has considered them the progenitors of the whole arthropodan phylum.

Specializations among the Arthropoda, even among the free-living forms, are so numerous that it would be difficult to make a complete list of them. In discussing the principal groups, I have tried to show that the essential structures can be explained as inherited from the Trilobita, changed in form by explainable modifications, and that new structures, not' present in the Trilobita, are of such a nature that they might be acquired independently in even unrelated groups.

The chief objections to the derivation of the remainder of the Crustacea from the trilobites have been: first, that the trilobites had broad pleural extensions; second, that they had a large pygidium; and lastly, that they had only one pair of tactile antennÆ.

It has now been pointed out that many modern Crustacea have pleural extensions, but that they usually bend down at the sides of the body, and also that in the trilobites and more especially in Marrella, there was a tendency toward the degeneration of the pleural lobes. A glance at the MesonacidÆ or ParadoxidÆ should be convincing proof that in some trilobites the pygidium is reduced to a very small plate.

In regard to the second antennÆ standard text-books contain statements which are actually surprising. A compilation shows that the antennÆ are entirely uniramous in but a very few suborders, chiefly among the Malacostraca; that they are biramous with both exopodite and endopodite well developed in most Copepoda, Ostracoda, and Branchiopoda; and that the exopodite, although reduced in size, still has a function in some suborders of the Malacostraca. The Crustacea could not possibly be derived from an ancestor with two pairs of uniramous antennÆ.

Although I have defended the trilobites, perhaps with some warmth, from the imputation that they were Arachnida, my argument does not apply in the opposite direction, and I believe Lankester was right in deriving the Arachnida from them. If the number of appendages in front of the mouth is fundamental, then the trilobites were generalized, primitive, and capable of giving rise to both' Crustacea and Arachnida. As shown on a previous page (p. 119), the "connecting links" so far found tend to disprove rather than to prove the thesis, but the present finds should be looked upon as only the harbingers of the greater ones which are sure to come.

LIMBS OF TRILOBITES PRIMITIVE.

The general presence, in an adult or larva, of some sort of biramous limbs throughout the whole class Crustacea has led most zoologists to expect such a limb in the most primitive crustaceans, and apparently the appendage of the trilobite satisfies the expectation. It is well, perhaps, as a test, to consider whether by modification this limb could produce the various types of limbs seen in other members of the class. In the first place, it is necessary to have clearly in mind the peculiarities of the appendage to be discussed.

It should first of all be remembered that the limb is articulated with the dorsal skeleton in a manner which is very peculiar for a crustacean. The coxopodite swings on a sort of ball-and-socket joint, and at the outer end both the exopodite and the basipodite articulate with it. Since the exopodite articulates with the basipodite as well as with the coxopodite, the two branches are closely connected with one another and there is little individual freedom of movement. This is, of course, a necessary consequence of their articulation with a segment which is itself too freely movable to provide a solid base for attachment of muscles. The relation of the appendifer, coxopodite, and two rami is here shown diagrammatically (fig. 33), the exopodite branching off from the proximal end of the basipodite at the junction with the coxopodite.

In all trilobites the endopodite consists of six segments, and the coxopodite of a single segment the inner end of which is prolonged as an endobase. There does not seem to be any variation from this plan in the subclass, although individual segments are variously modified. The exopodites are more variable, but all consist of a flattened shaft with setÆ on one margin. No other organs such as accessory gills, swimming plates, or brood pouches have yet been found attached to the appendages, the evidence for the existence of the various epipodites and exites described by Walcott being unsatisfactory (see p. 23).

In the Ostracoda the appendages are highly variable, but it is easily seen that they are modifications of a limb which is fundamentally biramous. In most species, both exopodite and endopodite suffer reduction. The exopodite springs from the basipodite and that segment is closely joined to the coxopodite, producing a protopodite. In some cases the original segments of the endopodites fuse to form a stiff rod. While highly diversified, these appendages are very trilobite-like, and some Ostracoda even have biramous antennÆ.

The non-parasitic Copepoda have limbs exceedingly like those of trilobites. Many of them are biramous, the endopodites sometimes retaining the primitive six segments. Coxopodite and basipodite are generally united, and endopodite and exopodite variously modified. Like some of the Ostracoda, the more primitive Copepoda have biramous antennÆ.

As would be expected, the appendages of the Cirripedia are much modified, although those of the nauplius are typical. The thoracic appendages of many are biramous, but both branches are multisegmented.

In the modern Malacostraca the ground plan of the appendages is biramous, but in most orders they are much modified. In many, however, the appendages of some part of the body are biramous, and in many the endopodites show the typical six segments. From the coxopodites arise epipodites, some of which assist in swimming, and some in respiration. Because of the many instances in which such extra growths arise, and because of the form of the appendages of the Branchiopoda, it has been suggested that the primitive crustacean leg must have been more complex than that of the trilobite. In looking over the Malacostraca, however, one is struck by the fact that epipodites generally arise where the exopodites have become aborted or are poorly developed, and seem largely to replace them. The coxopodite and basipodite are usually fused to form a protopodite, and a third segment is sometimes present in the proximal part of the appendage.

In the Branchiopoda are found the most complex crustacean limbs, and the ones most difficult to homologize with those of trilobites. In recent years, Lankester's homologies of the parts of the limbs of Apus with those of the Malacostraca have been quite generally accepted, and the appendages of the former considered primitive. Now that it is known that the Branchiopoda of the Middle Cambrian (Burgessia et at.) had simple trilobite-like appendages, it becomes necessary to exactly reverse the opinion in this matter. The same homologies stand, but the thoracic limbs of Apus must be looked upon as highly specialized instead of primitive.

Lankester (Jour. Micros. Sci., vol. 21, 1881) pointed out that the axial part of the thoracic limb of Apus (fig. 34) is homologous with the protopodite in the higher Crustacea, that the two terminal endites corresponded to the exopodite and endopodite, and that the other endites and exites were outgrowths from the protopodite analogous to the epipodites of Malacostraca. There seems to be no objection to retaining this interpretation, but with the meaning that both endopodite and exopodite are much reduced, and their functions transferred to numerous outgrowths of the protopodite. One of the endites grows inward to form an endobase, the whole limb showing an attempt to return to the ancestral condition of the trilobite. The limbs of some other branchiopods are not so easy to understand, but students of the Crustacea seem to have worked out a fairly satisfactory comparison between them and Apus.

The discovery that the ancestral Branchiopoda had simple biramous appendages instead of the rather complex phyllopodan type is another case in which the theory of "recapitulation" has proved to hold. It had already been observed that in ontogeny the biramous limb preceded the phyllopodan, but so strong has been the belief in the primitive character of the ApodidÆ that the obvious suggestion has been ignored. Even in such highly specialized Malacostraca as the hermit crabs the development of certain of the limbs illustrates the change from the schizopodal to the phyllopodan type, and Thompson (Proc. Boston Soc. Nat. Hist., vol. 31, 1903, pl. 5, fig. 12) has published an especially good series of drawings showing the first maxilliped. In the first to fourth zoeÆ the limb is biramous but in the glaucothoe a pair of broad processes grow out from the protopodite, while the exopodite and particularly the endopodite become greatly reduced. In the adult the endopodite is a mere vestige, while the flat outgrowths from the protopodite have become very large and bear setÆ.

Summary.

The limbs of most Crustacea are readily explained as modifications of a simple biramous type. These modifications usually take the form of reduction by the loss or fusion of segments and quite generally either the entire endopodite or exopodite is lacking. Modification by addition frequently occurs in the growth of epipodites, "endites," and "exites" from the coxopodite, basipodite, or both. A protopodite is generally formed by the fusion of coxopodite and basipodite, accompanied by a transference of the proximal end of the exopodite to the distal end of the basipodite. A new segment, not known in the trilobites (precoxal), is sometimes added at the inner end.

Among modern Crustacea, the anterior cephalic appendages and thoracic appendages of the Copepoda and the thoracic appendages of certain Malacostraca, Syncarida especially, are most nearly like those of the trilobite. The exact homology, segment for segment, between the walking legs of the trilobite and those of many of the Malacostraca, even the Decapoda, is a striking instance of retention of primitive characteristics in a specialized group, comparable to the retention of primitive appendages in man.

NUMBER OF SEGMENTS IN THE TRUNK.

Various attempts have been made to show that despite the great variability, trilobites do show a tendency toward a definite number of segments in the body.

Emmrich (1839), noting that those trilobites which had a long thorax usually had a short pygidium, and that the reverse also held true, formulated the law that the number of segments in the trunk was constant (20 + 1) Very numerous exceptions to this law were, however, soon discovered, and while the condition of those with less than twenty-one segments was easily explained, the increasing number of those with more than twenty-one soon brought the idea into total disrepute.

Quenstedt (1837) had considered the number of segments of at least specific importance, and both he and Burmeister (1843) considered that the number of segments in the thorax must be the same for all members of a genus. As first shown by Barrande (1852. p. 191 et seq.), there are very many genera in which there is considerable variation in the number of thoracic segments, and a few examples can be cited in which there is variation within a species, or at least in very closely related species.

Carpenter (1903, p. 333) has tabulated the number of trunk segments of such trilobites as were listed by Zittel in 1887 and finds a steady increase throughout the PalÆozoic. His table, which follows, is, however, based upon very few genera.

Due chiefly to the efforts of Walcott, an increasingly large number of Cambrian genera are now represented by entire specimens, and since these most ancient genera are of greatest importance, a few comments on them may be offered.

The total number of segments can be fairly accurately determined in at least nineteen genera of trilobites from the Lower Cambrian. These include eight genera of the MesonacidÆ (Olenellus was excluded) and Eodiscus, Goniodiscus, Protypus, Bathynotus, Atops, Olenopsis, Crepicephalus, Vanuxemella, Corynexochus, Bathyuriscus, and Poliella. The extremes of range in total segments of the trunk is seen in Eodiscus (9) and PÆdeumias (45+), and these same genera show the extremes in the number of thoracic segments, there being 3 in the one and 44+ in the other. PÆdeumias probably shows the greatest variation of any one genus of trilobites, various species showing from 19 to 44+ thoracic segments. The average for the nineteen genera is 13.9 segments in the thorax, 3.7 segments in the pygidium, or a total average of 17.6 segments in the trunk. Crepicephalus with 12-14 segments in the thorax and 4-6 in the pygidium, and Protypus, with 13 in the thorax and 4-6 in the pygidium, are the only genera which approach the average. All of the MesonacidÆ, except one, Olenelloides, have far more thoracic and fewer pygidial segments than the average, while the reverse is true of the EodiscidÆ, Vanuxemella, Corynexochus, Bathyuriscus, and Poliella.

The eight genera of the MesonacidÆ, Nevadia, Mesonacis, Elliptocephala, Callavia, Holmia, Wanneria, PÆdeumias, and Olenelloides, have an average of 20.25 segments in the thorax and 1.5 in the pygidium, a total of 21.75. If, however, the curious little Olenelloides be omitted, the average for the thorax rises to 22.14 and the total to 23.84. Olenelloides is, in fact, very probably the young of an Olenellus. Specimens are only 4.5 to 11 mm. long, and occur in the same strata with Olenellus (see Beecher 1897 A, p. 191).

Thirty-three genera from the Middle Cambrian afford data as to the number of segments, the AgnostidÆ being excluded. The extreme of variation there is smaller than in the Lower Cambrian. The number of thoracic segments varies from 2 in Pagetia to 25 in Acrocephalites, and these same genera show the greatest range in total number of trunk segments, 8 and 29 respectively.

The average of thoracic segments for the entire thirty-three genera is 10.5, of pygidial segments 5.9, a total average of 16.4. It will be noted that the thorax shows on the average less and the pygidium more segments than in the Lower Cambrian. If the AgnostidÆ could be included, this result would doubtless be still more striking. Of the genera considered, Asaphiscus with 7-11 thoracic and 5-8 pygidial segments, Blainia with 9 thoracic and 6-11 pygidial, Zacanthoides with 9 thoracic and 5 pygidial, and Anomocare with 11 thoracic and 7-8 pygidial segments came nearest to the average. Only a few departed widely from it. The genera tabulated were Acrocephalites, Alokistocare, Crepicephalus, Karlia, Hamburgia, Corynexochus, Bathyuriscus, Poliella, Agraulos, Dolichometopus, Ogygopsis, Orria, Asaphiscus, Neolenus, Burlingia, Blainia, Blountia, Marjumia, Pagetia, Eodiscus, Goniodiscus, Albertella, Oryctocara, Zacanthoides, Anomocare, Anomocarella, Coosia, Conocoryphe, Ctenocephalus, Paradoxides, Ptychoparia, Sao, and Ellipsocephalus.

Enough genera of Upper Cambrian trilobites are not known from entire specimens to furnish satisfactory data. Excluding from the list the Proparia recently described by Walcott, the average total trunk segments in ten genera is 18, but as most of the genera are OlenidÆ or olenid-like, not much weight can be attached to these figures.

For the Cambrian as a whole, the average for sixty-two genera is between 17 and 18 trunk segments, which is surprisingly like the result obtained by Carpenter from only twelve genera, and tends to indicate that it must be somewhere near the real average. If the 5 or 6 segments of the head be added, it appears that the "average" number of segments is very close to the malacostracan number 21. Genera with 16 to 18 trunk segments are Callavia, Protypus, Bathynotus, Crepicephalus, Bathyuriscus, Ogygopsis, Burlingia, Orria, Asaphiscus, Blainia, Zacanthoides, Neolenus, Anomocare, Conocoryphe, Saukia, Olenus, and Eurycare.

The order Proparia originated in the Cambrian, and Walcott has described four genera, one from the Middle, and three from the Upper. The number of segments in these genera is of interest. Burlingia, the oldest, has 14 segments in the thorax and 1 in the pygidium. Of the three genera in the Upper Cambrian, Norwoodia has 8-9 segments in the thorax and 3-4 in the pygidium; Millardia 23 in thorax and 3-4 in pygidium; and Menomonia 42 in thorax and 3-4 in pygidium. It is of considerable interest and importance to note that the very elongate ones are not from the Middle but from the Upper Cambrian.

Forty genera of Ordovician trilobites known from entire specimens were tabulated, and it was found that the range in the number of segments in the thorax and pygidium was surprisingly large. Agnostus, which was not included in the table, has the fewest, and Eoharpes, with 29, the most. While the range in number of segments in the thorax is 2 to 29, the range of the number in the pygidium, 2 to 26, is almost as great. A species of Dionide has 26 in the pygidium, while Remopleurides and Glaphurus have evidence of only 2. The average number of segments in the thorax for the forty genera was 10.15, in the pygidium 8.81, and the average number for the trunk 19.

Genera with just 19 segments in the trunk appear to be rare in the Ordovician, a species of Ampyx being the only one I have happened to notice. Calymene, Tretaspis, Triarthrus, Asaphus, Ogygites, and Goldius come with the range of 18 to 20. Goldius, with 10 segments in the thorax and (apparently) 8 in the pygidium, comes nearest to the averages for these two parts of the trunk. Goldius, Amphilichas, Bumastus, Acidaspis, Actinopeltis, and SphÆrexochus are among the genera having 10 segments in the thorax, and there are many genera which have only one or two segments more or less than 10.

In most Ordovician genera, thirty-five out of the forty tabulated, the number of segments in the thorax is fixed, and the variation is in any case small. In four of the five genera where it was not fixed, there was a variation of only one segment, and the greatest variation was in Pliomerops, where the number is from 15 to 19. This of course indicates that the number of segments in the thorax tends to become fixed in Ordovician time. The variation in the number of segments in the pygidium is, however, considerable. It is difficult in many cases to tell how many segments are actually present in this shield, as it is more or less smooth in a considerable number of genera. Extreme cases of variation within a genus are found in Encrinurus, species of which have from 7 to 22 segments in the pygidium, Cybeloides with 10 to 20, and Dionide with 10 to 26. As the number in the thorax became settled, the number in the pygidium became more unstable, so that not even in the Ordovician can the total number of segments in the trunk be said to show any tendency to become fixed.

The genera used in this tabulation were: Eoharpes, Cryptolithus, Tretaspis, Trinucleus, Dionide, Raphiophorus, Ampyx, Endymionia, Anisonotus, Triarthrus, Remopleurides, Bathyurus, Bathyurellus, Ogygiocaris, Asaphus, Ogygites, Isotelus, Goldius, Cyclopyge, Amphilichas, Odontopleura, Acidaspis, Glaphurus, Encrinurus, Cybele, Cybeloides, Ectenonotus, Calymene, Ceraurus, Pliomera, Pliomerops, Pterygometopus, Chasmops, Eccoptochile, Actinopeltis, SphÆrexochus, Placoparia, Pilekia, Selenopeltis, and Calocalymene.

Only sixteen genera of Devonian trilobites were available for tabulation, and it is not always possible to ascertain the exact number of segments in the pygidium, although genera with smooth caudal shields had nearly all disappeared. The number of segments in the thorax had become pretty well fixed by the beginning of the Devonian, Cyphaspis with a range of from 10 to 17 furnishing the only notable exception. The range for the sixteen genera is from 8 to 17, the average 11, the number exhibited by the PhacopidÆ which form so large a part of the trilobites of the Devonian. The greater part of the species have large pygidia, and while the range is from 3 to 23, the average is 11.2. Probolium, with 11 in the thorax and 11-13 in the pygidium, and Phacops, with 11 in the thorax and 9-12 in the pygidium, approach very closely to the "average" trilobite, and various species of other genera of the PhacopidÆ have the same number of segments as the norm. In every genus, however, the number of segments in the pygidium is variable, the greatest variation being in Dalmanites, with a range of from 9 to 23. The number of segments in the pygidium was therefore not fixed and was on the average higher than in earlier periods.

The genera used in the tabulation were: Calymene, Dipleura, Goldius, ProËtus, Cyphaspis, Acidaspis, Phacops, Hausmania, Coronura, Odontochile, Pleuracanthus, Calmonia, Pennaia, Dalmanites, Probolium, and Cordania.

The trilobites of the late PalÆozoic (Mississippian to Permian) belong, with two possible exceptions, to the PrÖetidÆ, and only three genera, ProËtus, Phillipsia, and Griffithides, appear to be known from all the parts. I am, however, assuming that both Brachymetopus and Anisopyge have 9 segments in the thorax, and so have tabulated five genera. The range in the number of segments in the pygidium is large, from 10 in some species of ProËtus to 30 in Anisopyge, and the average, 17.3, is high, as is the average for total number in the trunk, 26.3. Anisopyge, a late Permian trilobite described by Girty from Texas, is perhaps the last survivor of the group. It seems to have had 39 segments in the trunk, making it, next to the Cambrian PÆdeumias and Menomonia, the most numerously segmented of all the trilobites.

The above data may be summarized in the following table:

This table confirms that made up by Carpenter, and shows even more strikingly the progressive increase in the average number of segments in the trunk throughout the PalÆozoic.

While the two trilobites with the greatest number of segments are Cambrian, yet on the average, the last of the trilobites had the more numerously segmented bodies. The multisegmented trilobites are:

Anisopyge, the last of the trilobites, stands third on the list of those having great numbers of segments, and in each period there are a few which have considerably more than the average number. It may be of some significance that of these nine genera only PÆdeumias and Anisopyge belong to the Opisthoparia, the great central group, and that five are members of the Proparia, the latest and most specialized order.

FORM OF THE SIMPLEST PROTASPIS.

It would naturally be expected that the young of the Cambrian trilobites should be more primitive than the young of species from later formations, and Beecher (1895 C) has shown that this is the case. He had reference, however, chiefly to the eyes, free cheeks, and spines, and by comparison of ontogeny and phylogeny, demonstrated the greater simplicity of the protaspis which lacked these organs. It remains to inquire which among the other characteristics are most fundamental.

Among the trilobites of the Lower Cambrian, no very young have been seen except of MesonacidÆ. Of these, the ontogeny of Elliptocephala asaphoides Emmons is best known, thanks to Ford, Walcott, and Beecher, but, as the last-named has pointed out, the actual protaspis or earliest shield has not yet been found. The youngest specimen is the one roughly figured by Beecher (1895 C, p. 175, fig. 6). It lacks the pygidium, but if completed by a line which is the counterpart of the outline of the cephalon, it would have been 0.766 mm. long. The pygidium would have been 0.183 mm. long, or 23 per cent of the whole length. The axial lobe was narrow, of uniform width along the cephalon, showed a neck-ring and four indistinct annulations, but did not reach quite to the anterior end, there being a margin in front of the glabella about 0.1 mm. wide. The greatest width of the cephalon was 0.66 mm., and of the glabella 0.233 mm., or practically 35 per cent of the total width. Other young Elliptocephala up to a length of 1 mm., and young PÆdeumias, Mesonacis, and Holmia (see KiÆr, Videnskaps, Skrifter, 1 Mat.-Naturv. Klasse, 1917, No. 10) show about the same characteristics, but all these have large compound eyes on the dorsal surface and specimens in still younger stages are expected. It may be pointed out, however, that in these specimens the pygidium is proportionately larger than in the adult. Walcott cites one adult 126 mm. long in which the pygidium is 6 mm. long, or between 4 and 5 per cent of the total length, while in the incomplete specimen described above, it was apparently 23 per cent. In a specimen 1 mm. long figured by Walcott, the pygidium is 0.15 mm. long, or 15 per cent of the whole length.

The development of several species of trilobites from the Middle Cambrian is known. Barrande (1852) described the protaspis of Sao hirsuta, Peronopsis integer, Phalacroma bibullatum, P. nudum, and Condylopyge rex. Broegger figured that of a Liostracus (Geol. For. FÖrhandl., 1875, pl. 25, figs. 1-3) and Lindstroem (1901, p. 21) has reproduced the same. Matthew (Trans. Roy. Soc. Canada, vol. 5, 1888, pl. 4, pls. 1, 2) has described the protaspis of a Liostracus, Ptychoparia linnarssoni Broegger, and Solenopleura robbi Hartt. Beecher (1895 C, pl. 8) has figured the protaspis of Ptychoparia kingi Meek, and the writer that of a Paradoxides (Bull. Mus. Comp. Zool., vol. 58, No. 4, 1914, pl. i).

Sao, Liostracus, Ptychoparia, and Solenopleura all have the same sort of protaspis. In all, the axial lobe reaches the anterior margin and is somewhat expanded at that end; in all, the glabella shows but slight trace of segmentation; and in all, the pygidium occupies from one fifth to one fourth the total length. There is considerable variation in the width of the axial lobe. It is narrowest in Ptychoparia, where in the middle it is only 14 per cent of the whole width, and widest in Solenopleura, where it is 28 per cent. In Ptychoparia the pygidium of the protaspis occupies from 18 to 22 per cent of the whole length. In the adult it occupies 10 to 12 per cent. In Solenopleura it makes up about 26 per cent of the protaspis, and in the adult about 8 per cent.

In the youngest stages of all these trilobites, the pygidium is incompletely separated from the cephalon. The first sign of segmentation is a transverse crack which begins to separate the cephalon and pygidium, and by the time this has extended across the full width the neck segment has become rather well defined. In this stage the animal is prepared to swim by means of the pygidium, and first becomes active. The coincident development of the free pygidium and the neck-ring strongly suggests that the dorsal longitudinal muscles are attached beneath the neck-fur row.

The single protaspis of Paradoxides now known, while only 1 mm. long, is not in the youngest stage of development. It is like the protaspis of Olenellus in having large eyes on the dorsal surface and a narrow brim in front of the glabella. The glabella is narrower than in the adult.

The initial test of no agnostid has probably as yet been seen, as all the young now known show the cephalon and pygidium distinctly separated. Phalacroma bibullatum and P. nudum are both practically smooth and isopygous when 1.5 mm. long. P. bibullatum shows no axial lobe at this stage, but a wide glabella and median tubercle develop later, and when the glabella first appears, it extends to the anterior margin. In Peronopsis integer and Condylopyge rex, the axial lobe is outlined on each of the equal shields in specimens about 1 mm. long, but is without furrows and reaches neither anterior nor posterior margin.

From the foregoing brief description it appears that the pygidium of the protaspis varies in different groups from as little as 15 per cent of the total length in the MesonacidÆ to as much as 50 per cent in the AgnostidÆ; that the axial lobe varies from as little as 14 per cent of the total width in one Ptychoparia to as much as 50 per cent in Phalacroma nudum; that the glabella reaches the anterior margin in the OlenidÆ, SolenopleuridÆ, and Phalacroma bibullatum, while there is a brim in front of it in the OlenellidÆ, ParadoxidÆ, and three of the species of the AgnostidÆ. The decision as to which of these conditions are primitive may be settled quite satisfactorily by study of the ontogeny of the various species.

ORIGIN OF THE PYGIDIUM.

Taking first the pygidium, it has already been pointed out that in each case the pygidium of the adult is proportionally considerably smaller than the pygidium of the protaspis. The stages in the growth of the pygidium are better known in Sao hirsuta than in any other trilobite, and a review of Barrande's description will be advantageous.

Barrande recognized twenty stages in the development of this species, but there was evidently a still simpler protaspis in his hands than the smallest he figured, for he says, after describing the specimen in the first stage: "We possess one specimen on which the head extends from one border to the other of the disk, but as this individual is unique we have not thought it sufficient to establish a separate stage." This specimen is important as indicating a stage in which there was not even a suggestion of division between cephalon and pygidium.

In the first stage described by Barrande, the form is circular, the length is about 0.66 mm., and the glabella is narrow with parallel sides and no indications of lateral furrows. The neck segment is indicated by a slight prominence on the axial lobe, and back of it a constriction divides the axial lobe of the pygidium into two nodes, but does not cross the pleural lobes. The position of the nuchal segment permits a measurement of the part which is to form the pygidium, and shows that that shield made up 30 per cent of the entire length.

In the second stage, when the test is 0.75 mm. long, the cephalon and pygidium become distinctly separated, and the latter shield shows three annulations on the axial and two pairs of ribs on the pleural lobes. It now occupies 33-1/3 per cent of the total length.

In the third stage, when the total length is about 1 mm., the pygidium has continued to grow. It now shows five annulations on the axial lobe, and is 46 per cent of the total length.

In the fourth stage, two segments of the axial lobe have been set free from the front of the pygidium. The length is now 1.5 mm. and the pygidium makes up 32 per cent of the whole. From this time the pygidium continues to decrease in size in proportion to the total length, as shown in the following table.

This table shows the rapid increase in the length of the pygidium till the time when the thorax began to be freed, the very rapid decrease during the earlier part of its formation until six segments had been set free, and then a more gradual decrease until the entire seventeen segments had been acquired, after which time the relative length remained constant. From an initial proportion of 30 per cent, it rose to nearly one half the whole length, and then dwindled to a mere 6 per cent, showing conclusively that the thorax grew at the expense of the pygidium.

If this conclusion can be sustained by other trilobites, it indicates that the large pygidium is a more primitive characteristic of a protaspis than is a small one. I have already shown that the pygidium is proportionately larger in the protaspis in the MesonacidÆ, SolenopleuridÆ, and OlenidÆ, and a glance at Barrande's figures of "Hydrocephalus" carens and "H." saturnoides, both young of Paradoxides will show that the same process of development goes on in that genus as in Sao. There is first an enlargement of the pygidium to a maximum, a rise from 20 per cent to 33 per cent in the case of H. carens and then, with the introduction of thoracic segments, a very rapid falling off. All of these are, however, trilobites with small pygidia, and it has been a sort of axiom among palÆontologists that large pygidia were made up of a number of coalesced segments. While not definitely so stated, it has generally been taken to mean the joining together of segments once free. The asaphid, for instance, has been thought of as descended from some trilobite with rich segmentation, and a body-form like that of a Mesonacis or Paradoxides.

The appeal to the ontogeny does not give as full an answer to this question as could be wished, for the complete life-history of no trilobite with a large pygidium is yet known. While the answer is not complete, enough can be gained from the study of the ontogeny of Dalmanites and Cyclopyge to show that in these genera also the thorax grows by the breaking down of the pygidium and that no segment is ever added from the thorax to the pygidium. The case of Dalmanites socialis as described by Barrande (1852, p. 552, pl. 26) will be taken up first, as the more complete. The youngest specimen of this species yet found is 0.75 mm. long, the pygidium is distinctly separated from the cephalon, and makes up 25 per cent of the length. This is probably not the form of the shell as it leaves the egg. At this stage there are two segments in the pygidium, but they increase to four when the test is 1 mm. long. The cephalon has also increased in length, however, so that the proportional length is the same. The subjoined table, which is that compiled by Barrande with the proportional length of the pygidium added, is not as complete as could be desired, but affords a very interesting history of the growth of the caudal shield. The maximum proportional length is reached before the introduction of thoracic segments, and during the appearance of the first five segments the size of the pygidium drops from 25 to 15 per cent. Several stages are missing at the critical time between stages 8 and 9 when the pygidium had added three segments to itself and has supplied only one to the thorax. This would appear to have been a sort of resting or recuperative stage for the pygidium, for it increased its own length to 20 per cent, but from this stage up to stage 12 it continued to give up segments to the thorax and lose in length itself. After stage 12, when the specimens were 8 mm. long, no more thoracic segments were added, but new ones were introduced into the pygidium, until it reached a size equal to one fifth the entire length, as compared with one fourth in the protaspis.

Since the above was written, Troedsson (1918, p. 57) has described the development of Dalmanites eucentrus, a species found in the Brachiopod shales (Upper Ordovician) of southern Sweden. This species follows a course similar to that of D. socialis, so that the full series of stages need not be described. The pygidium is, however, of especial interest, for there is a stage in which it shows two more segments than in the adult. Troedsson figures a pygidium 1.28 mm. long which has eight pairs of pleural ribs, while the adult has only six pairs. The ends of all these ribs are free spines, and were the development not known one would say that this was a case of incipient fusion, while as a matter of fact, it is incipient freedom.

A further interest attaches to this case, because of the close relationship between D. eucentrus and D. mucronatus. The latter species appears first in the Staurocephalus beds which underlie the Brachiopod shales, so that in its first appearance it is somewhat the older. The pygidium of the adult D. mucronatus is larger than that of D. eucentrus, having eight pairs of pleural ribs, the same number as in the young of the latter. In short, D. eucentrus is probably descended from D. mucronatus, and in its youth passes through a stage in which it has a large pygidium like that species. Once more it appears that the small pygidium is more specialized than the large one.

The full ontogeny of Cyclopyge is not known, but young specimens show conclusively that segments are not transferred from the thorax to the pygidium, but that the opposite occurs. As shown by Barrande (1852) and corroborated by specimens in the Museum of Comparative Zoology, the process is as follows: The third segment of the adult of this species, that is, the fourth from the pygidium, bears a pair of conspicuous cavities on the axial portion. In a young specimen, 7 mm. long, the second segment bears these cavities, but as the thorax has only four segments, this segment is also the second instead of the fourth ahead of the pygidium. The pygidium itself, instead of being entirely smooth, as in the adult state, is smooth on the posterior half, but on the anterior portion has two well formed but still connected segments, the anterior one being more perfect than the other. These are evidently the two missing segments of the thorax, and instead of being in the process of being incorporated in the pygidium, they are in fact about to be cast off from it to become free thoracic segments. In other words, the thorax grows through the degeneration of the pygidium. That the thorax grows at actual expense to the pygidium is shown by the proportions of this specimen. In an adult of this species the pygidium, thorax, and cephalon are to each other as 9:11:13. In the young specimen they are as 10:6:12, the pygidium being longer in proportion both to the thorax and to the cephalon than it would be in the adult.

This conception of the breaking down of the pygidium to form the thorax will be very helpful in explaining many things which have hitherto seemed anomalous. For instance, it indicates that the AgnostidÆ, whose subequal shields in early stages have been a puzzle, are really primitive forms whose pygidia do not degenerate; likewise the EodiscidÆ, which, however, show within the family a tendency to free some of the segments. The annelidan MesonacidÆ may not be so primitive after all, and their specialized cephala may be more truly indicative of their status than has previously been supposed.

The facts of ontogeny of trilobites with both small and large pygidia do show that there is a reduction of the relative size of the caudal shield during the growth-stages, and therefore that the large pygidium in the protaspis is probably primitive. The same study also shows that the large pygidium is made up of "coalesced segments" only to the extent that they are potentially free, and not in the sense of fused segments.

WIDTH OF THE AXIAL LOBE.

That the narrow type of axial lobe is more primitive than the wide one has already been demonstrated by the ontogeny of various species, and space need not be taken here to discuss the question. Most Cambrian trilobites have narrow axial lobes even in the adult so that their development does not bring this out very strikingly, though it can be seen in Sao, Ptychoparia, etc., but in Ordovician trilobites such as Triarthrus and especially Isotelus, it is a conspicuous feature.

PRESENCE OR ABSENCE OF A "BRIM."

That the extension of the glabella to the front of the cephalon is a primitive feature is well shown by the development of Sao (Barrande, 1852, pl. 7), Ptychoparia (Beecher, 1895 C, pl. 8), and Paradoxides (Raymond, Bull. Mus. Comp. Zool., vol. 57, 1914), although in the last genus the protaspis has a very narrow brim, the larva during the stages of introduction of new segments a fairly wide one, and most adults a narrow one.

The brim of Sao seems to be formed partly by new growth and partly at the expense of the frontal lobe, for that lobe is proportionately shorter in the adult than in the protaspis. In Cryptolithus and probably in Harpes, Harpides, etc., the brim is quite obviously new growth and has nothing to do with the vital organs. Its presence or absence may not have any great significance, but when the glabella extends to the frontal margin, it certainly suggests a more anterior position of certain organs. In Sao, the only trilobite in which anything is known of the position of the hypostoma in the young, the posterior end is considerably further forward in a specimen a. 5 mm. long than in one 4 mm. long, thus indicating a backward movement of the mouth during growth, comparable to the backward movement of the eyes.

SEGMENTATION OF THE GLABELLA.

The very smallest specimens of Sao show a simple, unsegmented axial lobe, and the same simplicity has been noted in the young of other genera. Beecher considered this as due to imperfect preservation of the exceedingly small shells, which practically always occur as moulds or casts in soft shale. There is, however, a very general increase in the strength of glabellar segmentation in the early part of the ontogeny of all trilobites whose life history is known, and in some genera, like the AgnostidÆ, there is no question of the comparatively late acquisition of glabellar furrows. Even in Paradoxides, the furrows appear late in the ontogeny.

Summary.

If absence of eyes on the dorsal surface be primitive, as Beecher has shown, and if the large pygidium, narrow axial lobe, and long unsegmented glabella be primitive, then the known protaspis of the MesonacidÆ and ParadoxidÆ is not primitive, that of the OlenidÆ is very primitive, and that of the AgnostidÆ is primitive except that in one group the axial lobe, when it appears, is rather wide, and in the other a brim is present.

Subsequent development from the simple unsegmented protaspis would appear to show, first, an adaptation to swimming by the use of the pygidium; next, the invagination of the appendifers as shown in the segmentation of the axial lobe indicates the functioning of the appendages as swimming legs; then with the introduction of thoracic segments the assumption of a bottom-crawling habit is indicated. Some trilobites were fully adapted for bottom life, and the pygidium became reduced to a mere vestige in the production of a worm-like body. Other trilobites retained their swimming habits, coupled with the crawling mode of life, and kept or even increased (Isotelus) the large pygidium.