Numerical variations in the joints of reproduced appendages have been described by several observers in other Orthoptera than the Blattide and in certain other groups of the tracheate Arthropoda. As has been pointed out, the phenomena of reproduction of the legs and the special features of the peculiar form of tarsus associated therewith in the Cockroach must be considered with some reference to what is known of the reproduction of lost parts in the Arthropoda generally. The close connection between the phenomena of reproduction of appendages and ecdysis permits the insertion at this place of certain facts noticed during the mutilation experiments already briefly described.

Immediate effects of mutilation.-A drop of blood appeared on the cut or ruptured surface of the leg, but clotted in a minute or two, thus preventing further hæmorrhage. The loss of part or the whole of the leg seemed to inflict mechanical inconvenience only, and an individual which had lost portions of three legs moved about on recovery from anæsthesia with fair activity and resumed its normal habits at once. It was noticed that in nearly all cases the remaining portions of a partially removed tarsus were very soon dropped off, the tibia then terminating the limb. The remaining half of a divided tibia was sometimes dropped and sometimes retained. These losses of parts proximal to the artificially injured region never occurred simultaneously with the inflicted injury. In considering this matter it is necessary to bear in mind the fact that Arthropods of several groups have the power of throwing off their appendages in response to stimuli of various kinds, a phenomenon to which the name autotomy has been given. In Cockroaches there seems to exist a very slight degree of autotomy. If thrown into boiling water they do not snap off their legs as is the case, for instance, with many Spiders. On the other hand, a Cockroach held by a leg not infrequently escapes by its separation from the body when no particularly strong pull is made by the forceps holding the limb, and with a suddenness suggestive of autotomy. Moreover, it was noticed that the break occurs, invariably, either at the tarso-tibial articulation or (and much more frequently) at the suture where femur and trochanter are fused. But if a certain degree of autotomy be admitted, it must be remembered that, in a Cockroach preserved in spirit, a break is effected easily at either of the above-mentioned places, while a strong pull with the forceps is necessary to separate femur and tibia. If the body be held and the tarsus pulled, the break occurs at the femoro-trochanteric suture, while if the femur be held instead, a pull on the tarsus is followed by its separation from the tibia.

I cannot find any record of observation on this point in the Blattidæ, but in the case of the Phasmidæ, Scudder1 observed that in Diapheromera amputation of any portion of a leg distal to the femoro-trochanteric suture was followed by loss before the next

1 Proc. Boston Nat. Hist. Soc. 1869, xii. p. 99.

ecdysis of the remaining parts up to the suture. Scudder does not speak of immediate loss suggesting autotomy as usually understood, but he states that the regrowth which replaced the lost parts necessarily always commenced from the above place. On the other hand, Bordage describes well-marked autotomy and subsequent reproduction in two other genera of Phasmids as always occurring at the femoro-trochanteric suture. As in the case of the Blattiaa, these observations on the Phasmide were on individuals which had not performed their final ecdysis, after which reproduction of lost appendages ceases.

3

Bordage, who employed different kinds of stimuli for bringing about autotomy, found that the age of the individual, as well as the mode of stimulation, was a controlling factor in the readiness with which autotomy occurred. Heineken 2, early in this century, experimented on the autotomy of the posterior, or jumping, legs in genera of Gryllidæ, Locustidæ, and Acridiida; while more recently Frédéricq and Contejean have observed in detail the autotomy of the jumping-legs of Locusta viridissima. The experiments of these authors show that the autotomy of the jumping-legs takes place at the femoro-trochanteric suture. [The statement of the last-named that inasmuch as the trochanter is absent in Locusta viridissima, the autotomous break occurs between femur and coxa, appears to rest on the fact that in this species the trochanter is telescoped into the coxa in such a manner as to be visible only when the femur is removed. In a spirit-specimen a sharp pull on the femur always leaves the trochanter still attached to the coxa. In connection with this point, as well as with others in the present enquiry, I am indebted to Dr. David Sharp, F.R.S., for much kind advice and assistance.]

The above-mentioned authors, in addition to ascertaining that different methods of injury and stimulation caused autotomy after a shorter or longer latent period, demonstrated that the event is dependent on the integrity of the third pair of thoracic ganglia, and is as truly a reflex action as the autotomy exhibited by the appendages of Decapod Crustacea, or the tail in certain Lizards. In connection with these observations it must be noted that the power of reproducing lost legs is usually supposed to be absent in the Orthoptera Saltatoria. Graber has observed reproduction of the antennæ in Gryllus and Locusta, but could not obtain reproduction of the tarsi. For further information the writings

5

1 "Phénomènes d'autotomie observés chez les nymphes de Monandroptera inuncans et de Rhaphiderus scabrosus," Comptes-Rendus Acad. Sci. 1897, cxxiv. pp. 210 & 378.

2 Loc. cit. p. 427.

3 "Les mutilations spontanées ou l'autotomie," Rev. Scientifique, 1886, sér. 3, xii. p. 613.

4.66 Sur l'autotomie chez la Sauterelle et le Lézard," Compt.-Rend. Acad. Sci. 1890, cxi. p. 611.

5 Zur Entwickelungsgeschichte und Reproductionsfähigheit der Orthopteren," Sitzungsb. der Akad. d. Wiss., Math.-naturw. Cl. lv. Bd. i., Wien, 1867; also Ann. & Mag. Nat. Hist. ser. 3, xix. p. 147.

4

of the above-mentioned authors and those of Durieu ', Frédéricq, Peyerimhoff, and Werner may be referred to. More recently, however, Griffini, quoting the above in connection with cases he has observed of apparent reproduction of appendages in Gomphocerus, Oedipoda, and Pristes, concludes that, as is the case in the Cursoria, reproduction may not improbably occur during the post-embryonic development of the Saltatoria also.

From these statements it appears that more extended observations would show that among the Orthoptera alone the more immediate effects of injury to a limb differ considerably in the several tribes of the order. While in some genera a slight stimulus may bring about immediate autotomy at a certain fixed place, in others little or no autotomy is observable even when strong means of stimulation, such as amputation of the more distal portions of a limb, are employed, the injury in such cases being followed sometimes by the eventual dropping away of certain portions of the remaining stump, and sometimes by the retention of the entire stump. Again, though the subject of autotomy necessarily bears a close relation to that of reproduction of lost parts, there seems to be no direct ratio between the degree of autotomy exhibited and the power of reproduction possessed in any particular case.

More or less parallel are the diverse results obtained from mutilation of the limbs in different genera of Arachnida recorded by Heineken, Black wall', Parize, and Frédéricq, to whose work fuller reference will be made later on in connection with the phenomena of reproduction of the lost parts.

Mortality in confinement.-About 25% of the 833 mutilated individuals died before accomplishing an ecdysis. Apparently this mortality was not due to the injuries inflicted, for it was not excessive in the period immediately following mutilation, but occurred at a steady rate throughout the experiment. The animals were kept in three glass-fronted boxes 24 x 8 x 10 inches in size, and provided with narrow dark shelters imitating the crevices haunted by Cockroaches when at large. The boxes remained throughout the experiment in a room kept at a temperature of 16° C. It is very possible that these arrangements reproduced the natural habitats too imperfectly and that overcrowding, or want

1 "Notes sur quelques Orthoptères," Petites Nouv. Entomol. 1876, no. 158. 2 Loc. cit.; also 'La Lutte pour l'existence chez les Animaux Marins' (Paris, 1889), p. 259.

3

Note sur l'atrophie des membres chez les Orthoptères," Miscellanea Entomologica, 1896, iv. p. 70.

4 66 Selbstverstümmelung bei Heuschrecken," Zool. Anzeiger, 1892, Jahrg. xv. p. 58.

5 "Di un Pristes tuberosus anomalo," Boll. de Musei di Zool. ed Anat. Comp. Torino, 1896, xi. no. 234; also "Di due Acrididi anomali," ibid. xi. no. 256. 6 Loc. cit.

7.66 Report on some recent Researches into the Structure, Functions, and Economy of the Araneidea made in Great Britain," Brit. Assoc. Rep. 1845, p. 62.

"L'amputation réflexe des pattes des Crustacés," Rev. Scientifique, 1886, sér. 3, xi. p. 379. 9 Loc. cit.

of sufficient space for running about, was the chief cause of this degree of mortality. That it was abnormally high seems probable, especially in view of the facts which will be mentioned later on, concerning the hatching of the egg-cases. Moreover, though Cockroaches once established in a building may become extremely numerous, their slowness in spreading in the neighbourhood is well known, and it is probable that we by no means understand what conditions are most favourable to this imported species. I succeeded, however, in maintaining a few individuals in health for nearly two years and one lived for rather over that time, and eventually escaped from confinement. Bread and cake were always readily eaten, but the animals seemed to pay no attention to a moist sponge always kept in the cages. With large numbers in one cage it was impossible to observe how often any one individual made a meal, but Dr. Sharp' has called attention to the very small amount of food that is required to maintain a Cockroach in good condition, and that many weeks of starvation seem to make little difference to the animal. In the case of my own Cockroaches, a weak individual was never attacked by his companions, but the soft parts of the dead were soon devoured, as were all cast skins not removed from the cages. These latter were certainly not always eaten by their owners, though occasionally they were.

6

Occurrence of the Ecdyses.-On this subject the monograph of Miall and Denny' contains the following statement in quotation of Cornelius :-"The first change of skin occurs immediately after escape from the egg-capsule, the second four weeks later, the third at the end of the first year, and each succeeding moult after a year's interval. At the sixth month the insect becomes a pupa,' and at the seventh (being now four years old) it assumes the form of the perfect insect. The changes of skin are annual and, like fertilization and oviposition, take place in the summer months only. These statements are partly based on observations of captive Cockroaches, and are the only ones accessible; but they require confirmation by independent observers, especially as they altogether differ from Hummel's account of the life-history of Blatta germanica, and are at variance with the popular belief that new generations of the Cockroach are produced with great rapidity."

4

The observation of Hummel referred to is that Phyllodromia germanica performs six ecdyses between April and September, and becomes adult within seven months from hatching out of the egg-capsule. My own observations on the post-embryonic development of S. orientalis may now be described.

In the first place great difficulty was experienced in obtaining

1 Cambr. Nat. Hist. v. P. 229.

2 The Structure and Life-history of the Cockroach (Periplaneta orientalis)' (London, 1886), p. 23.

3 Beiträge zur näheren Kenntniss von Periplaneta orientalis. Elberfeld,

1853.

4

Quelques observations sur la Blatte germanique."

giques. I. St. Pétersbourg, 1821.

Essais Entomolo

undoubtedly newly hatched individuals. Between 500 and 600 egg-capsules were collected from bakehouses and kitchens in Cambridge, very many of them immediately after deposition by

the females.

At first they were placed in an incubator in the Zoological Laboratory in a constant temperature of 38° C., but so little success attended this method that the capsules obtained later were left undisturbed, as nearly as possible in the exact spots where they had been deposited. But from the whole number of capsules observed, only 20 hatchings were obtained, the total number of young thus raised being 210, an average of 10.5 young from each capsule, which in this species normally contains 16 embryos.

If this observation may be regarded as resting on a fairly sufficient number of instances, it would seem that a large proportion of the egg-cases deposited do not hatch out, and also that some of the young in those that do hatch never appear. As all the cases that hatched did so within a few days after deposition, while the others examined long after were invariably found to be quite brittle and dried up inside, it seems unlikely that the non-hatching of the greater number observed was merely a result of allowing insufficient time for hatching to take place. If this view be accepted, it affords a partial explanation of the well-known tardiness with which this species spreads.

With regard to the time of year when ecdysis occurred, the following observations were made. The mutilated Cockroaches commenced living in captivity during the first three months of the year. The dates of 235 observed ecdyses were distributed

That the number of ecdyses during the maximum period increased rather more rapidly than it declined, is apparent from the following: