Text-fig. 23.—Figs. 128–132—Anapistula australia n.sp. Fig. 128—Dorsal surface of carapace. Fig. 129—Chelicera of female. Fig. 130—Colulus. Fig. 131—Female epigynum. Fig. 132—Internal genitalia.

studied and used as a basic character in an attempt to elucidate phylogeny within the order. In 1933 Professor Alexander Petrunkevitch published the results of an extensive survey of both the internal and external structure of the spiders and set out an overall classification in keeping with his findings. The Symphytognathidae he placed with the Caponiidae and Telemidae into a separate suborder, the Apneumonomorphae. Subsequent authors (Bristowe, 1938; Fage, 1937) have disagreed with this conclusion, maintain that the apneumone families do not form a natural assemblage and group the families with dipneumone families.

The sub-order Apneumonomorphae is based primarily on two internal characters which all these spiders have in common. These are the absence of anterior lungbooks and their replacement with tracheae, and the presence of only two pairs of ostia. The reduction in number of ostia is also found in other families of spiders leaving the absence of lungbooks and their replacement with trachea the only character unique to the three families. It would therefore seem that the only justification for a separate suborder would be if it could be postulated that the loss of the

anterior pair of lungbooks and their replacement with trachea happened only once during the common development of these families. Recent work by Fage and Machado (1951) and Machado (1951) on the Ochyroceratidae has demonstrated that in this family there are genera with a pair of anterior lungbooks and others which are without lungbooks. An examination of the respiratory system of the spiders studied in the present paper in conjunction with the results published by other authors convinces me that a similar situation exists in the Symphytognathidae and that there is little justification other than convenience for placing this family into a separate suborder with the Caponiidae and Telemidae. The difficulty expressed by Petrunkevitch in imagining a sudden change from functional lungbooks to equally functional tracheae is hard to understand. The functioning of the lungbooks and the tracheae of spiders, from the information available, would appear to be similar if not identical. Oxygen transfer is achieved through the walls of these structures to oxygenophylic bodies, which then carry the oxygen to the structure requiring it, a condition in contrast to insects where oxygen transfer is considered to take place directly from the tracheoles to the tissue and is correlated with the absence of oxygen carrying pigment in the blood. The gradual elongation of the lamellae of lungbooks and reduction in the width of the lumen would lead directly to a structure having the form and presumably the function of a tracheal system without any question arising of a hiatus in the efficient use of these structures in the respiration of the spider. That this is in fact what has happened is, I feel, indicated by the structure of the anterior respiratory system of Risdonius, Archerius and some species of Mysmena (Levi, 1956) where it is difficult to decide on morphological grounds whether the structures present should be termed modified lungbooks or a tracheal system. It therefore seems reasonable to assume that the change from lungbooks to tracheae has taken place a number of times and that the change is governed by physiological factors.

The tendency for lungbooks to be replaced by tracheae is of considerable general interest. Davies and Edney (1952) during their study on the evaporation of water from spiders demonstrated that in Lycosa amenta respiration took place mainly through the lungbooks, and that tracheal respiration alone was not sufficient to keep the spider alive. If, as might be concluded from these experiments that the lungbooks are the more efficient respiratory organ, it seems surprising that the overall evolution of the spiders indicates a progressive loss of these structures. It is perhaps significant that all of these spiders which we know have lost the anterior pair of lungbooks are small and with the exception of the Caponiidae are in fact minute. It is probable that with the reduction in size and the increase of surface area in relation to body volume, water loss becomes an increasingly important factor influencing changes in the respiratory system because the water loss from lungbooks could be much greater than from tracheae. Most of these spiders are found only in habitats where there is a constant high humidity and are difficult to keep under laboratory conditions for this reason. Furthermore, there is a tendency for many of them to possess sclerotic thickenings and plates on the abdomen which possibly reduce transpiration through the integument.

The respiratory system now known for the species placed in the Symphytognathidae covers a wide range, with a certain degree of uniformity at a generic level. If, as is most probable, the loss of a structure such as the posterior tracheae, the fusion of two spiracles into a single median one, or the change from lungbooks to tracheae, precludes the future reappearance of these structures in their earlier form, it is necessary to postulate an ancestral form which possessed one pair of lungbooks and two posterior spiracles leading into tracheae. The only living spiders which possess this arrangement are those placed in the families Dysderidae and Oonopidae, neither of which show very close relationship when other characters are considered. If, as is suggested in the present paper, these spiders have developed from the Argiopidae or as a number of other authors have suggested, the Theridiidae, then

we must look for forms within these families which still possess a pair of posterior spiracles. As far as I am aware none has been recorded but this does not preclude the actual existence of such forms either among the smaller known species or in forms at present not known. It would, however, be in no way surprising if this character has in fact been completely lost since the divergence of the Symphytognathidae from the parent stock in view of the number of forms this system takes within the Symphytognathidae and the overall indication that there is a tendency for the two posterior spiracles to merge into one.

Mysmena appears to have retained the primitive arrangement more consistently than other genera. Mysmena guttata (Banks) and Mysmena phyllicola (Marples) possess modified lungbooks, while the two posterior spiracles open into tracheae. In M. incredula (Gertsch and Davis), M. woodwardt n.sp. (Fig. 147), M. rotunda (Marples) and M. samoensis (Marples) (Fig. 148a) the anterior spiracles lead into tracheae which are discrete in woodwardt but joined by a transverse duct in other species. The two posterior spiracles in all these species lead into atria which are

Text-fig. 25.—Figs. 142–148—Respiratory systems (anterior tracheae stippled). Fig. 142—Chasmocephalon armatum, Forster. Fig. 143—Chasmocephalon minutum Hickman (from Hickman, 1944). Fig. 144—Symphytognatha globosa Hickman, female (from Hickman, 1931). Fig. 145—Chasmocephalon sp. ? Capetown (from Fage, 1937). Fig. 146—Patu Marples, female. Fig. 147—Mysmena woodwardt n.sp. immature female. Fig. 148—Mysmena vitiensis n.sp., female. Fig. 148A—Mysmena samoensis (Marples).

connected transversely and from each atrium tracheal tubes run through into the cephalothorax. In Mysmena vitiensis (Fig. 148), however, the anterior atria are not joined, and there appears to be a single median spiracle midway between the spinnerets which leads into four tubes, which do not enter the cephalothorax. The general pattern found in Patu is that illustrated for P. vitiensis (Fig. 146). The anterior spiracles lead into large atria which are connected transversely and tracheae from these atria supply both the cephalothorax and the abdomen. This is similar to the system found in Symphytognatha (Fig. 144). However, an undescribed species from Poutasi, in Western Samoa, which is undoubtedly a typical Patu, possesses two spiracles which are midway between the spinnerets and the epigastric groove as in most species of Mysmena. These lead into short atria which are joined transversely, and from each atrium numerous tracheae pass directly through to the cephalothorax as they do in Mysmena. The anterior spiracles in Risdonius lead into tubular atria which are not connected transversely; from the inner surface of these tubes extend a number of evenly spaced structures which in R. parvus Hickman (Fig. 133) have the appearance of modified lamellae and have been described as such by Hickman (1939), but which in R. conicum (Forster) (Fig. 134) are more elongate and tubular and have more the appearance of tracheae. In both of these species there is a single posterior spiracle at the base of spinnerets which opens into a short atrium from which runs three or four tracheae limited to the abdomen. In Chasmocephalon the anterior spiracles open into short atria from which tracheae are supplied to both the abdomen and the cephalothorax. In C. minutum Hickman (Fig. 143) and an undescribed species from Capetown examined by Fage (Fig. 145) there is no posterior spiracle, but in the New Zealand species C. armatum (Forster) (Fig. 142) the posterior spiracle is present at the base of the spinnerets and this leads into four tracheal tubes which are limited to the abdomen.

The tracheal system of Anapistula (Fig. 158) is very similar to that found in Mysmena. There are two pairs of spiracles; the anterior pair lead into short atria which are connected by a transverse tube, from each atrium five or six tubes extend throughout the abdomen. The posterior pair of spiracles are situated midway between the epigastric groove and the spinnerets and lead into short atria which are connected by a transverse duct while a thick bunch of tracheae run from each atrium directly to the cephalothorax.

Fage (1937) examined the respiratory system of Anapis hamigera (Simon) and found that the single posterior spiracle which is placed between the spinnerets and the epigastric groove leads into a short vestibule from which runs two pairs of large trunks. The numerous fine tracheae from these trunks were limited to the abdomen. The two anterior spiracles lead into a wide transverse vestibule which was broken up at each outer margin into two trunks passing through the petiolus to the cephalothorax. In Anapis mexicana Forster (Fig. 157) the position of the spiracles is the same, but the posterior spiracle is present as a broad slit which leads into a short atrium from which two bunches of tracheae lead directly into the cephalothorax while the tracheae from the anterior spiracles are limited to the abdomen. The system for Anapisona gertschi Forster (Fig. 155) is similar to A. mexicana Forster except that the posterior spiracle is placed at the base of the spinnerets. In Pseudanapis only the anterior spiracles are present. Pseudanapis algerica Simon (Fig. 156), P. relicta Kratochvil (Fage, 1937), P. octocula n.sp. (Fig. 152), P. burra n.sp., P. insula (Forster) (Fig. 153), and P. wilsont n.sp. (Fig. 154) all have bunches of tracheae passing through the petiolus to the cephalothorax, but in P. darlingtoni n.sp. (Fig. 150) and P. spinipes (Forster) (Fig. 151) the tracheae are limited to the abdomen and the atria are very long and tubular. In both of these latter two species the spiracles have moved anteriorly and open near the petiolus. There is a transverse connecting duct present in P. algerica, P. relicta and P. insula, but this duct is absent from all other species examined

Lucharachne palpalis Krauss (Fig. 149) is also without lungbooks and the system for this species is almost identical with that of Anapis hamigera, with the anterior spiracles providing tracheae to the cephalothorax as well as the abdomen. The posterior spiracle, however, is situated at the base of the spinnerets, and appears to have two small openings placed very close to each other, which open into a common atrium. In Micropholcomma (Figs. 135, 136, 137) the two anterior spiracles lead into large atria which are joined by a transverse duct. From the atria a number of trunks are limited to the abdomen, but a single pair pass through the petiolus and branch into numerous fine tracheae in the cephalothorax. The single posterior spiracle is situated at the base of the spinnerets and leads into a short atrium from which runs two or four short tracheae. Pua and Parapua are without a posterior spiracle, but the anterior tracheal system in Pua novaezealandiae n.sp. (Fig. 138) is the same as in Micropholcomma except that the transverse duct is absent. In Parapua punctata n.sp. (Fig. 141) a bunch of five fine tracheae pass through the petiolus in place of the single trunk in the other two genera. Textricella (Figs. 139, 140) is also without a posterior spiracle, but the tracheae from anterior spiracles are limited to the abdomen.

The overall picture appears to be one of active change in the form of the respiratory system within the family at a generic level. Only in Mysmena does there appear the original arrangement with two posterior spiracles leading into tracheae and two anterior spiracles leading into lungbooks and even in these species the lungbooks are not typical of other spiders. The changes appear to follow a fairly set pattern with, first, the modification of the anterior lungbooks into tracheae, then the fusion of the two posterior spiracles into a single median spiracle, which is then situated posteriorly at the base of the spinnerets in contrast to the placing of the two original spiracles, which are usually situated midway between the epigastric groove and the spinnerets. An intermediate stage is illustrated in Lucharachne where the posterior tracheae open from the base of the spinnerets through two openings placed on a common plate. The ultimate form is found when the posterior spiracle is lacking leaving the two anterior spiracles leading into tracheae as the sole respiratory organ. At this stage there appears to be a tendency for the spiracles to move anteriorly beyond the epigastric groove as in Pseudanapis spinipes and P. darlingtoni.

In most genera tracheae are supplied to the cephalothorax from either the anterior or posterior spiracles, and in no case have tracheae been recorded penetrating to the cephalothorax from both. In eight of the genera examined (Symphytognatha, Patu, Micropholcomma, Pua, Parapua, Chasmocephalon, Pseudanapis and Lucharachne) tracheae from the anterior pair of spiracles supply both the abdomen and the cephalothorax, but in two species of Pseudanapis (spinipes and darlingtoni) the anterior tracheae are limited to the abdomen. In four of the genera (Mysmena, Anapistula and Anapisona) tracheae are supplied to the cephalothorax from the posterior spiracles and except in Anapisona all of the trachea pass directly through the petiolus to the cephalothorax. In only two genera (Risdonius and Textricella) are tracheae not present in the cephalothorax. The presence in some species of a transverse tube connecting the atria is interesting, and may be found to have some significance, although from the data available at present this does not appear to be so. A similar duct joins the lungbooks of many dipneumone spiders.

The distribution and origin of the tracheae to the abdomen and cephalothorax has been used as a major character in the separation of the families previously placed in the Apneumonomorphae, but it now seems that this has little significance beyond a generic level and can vary within a genus.