Brunner' quotes Serville's remarks, and states that he could add a large number of cases. He observes that the abnormal tarsus is usually unilateral and on the posterior legs (Nyctibora, Epilampra, &c.).

The above-mentioned authors examined only small numbers of individuals, and it was therefore desirable to extend the inquiry as to the frequency of occurrence of abnormal tarsi through greater numbers. This has been done among certain easily obtained species, the total numbers examined being :

:

The specimens of P. americana and P. germanica were obtained from the Society's Gardens, those of P. australasia from the Botanic Garden, Cambridge (where the egg-cases of this species appear to have been introduced in packets of plants from Kew and South America), and those of S. orientalis from Cambridge bakehouses, except 262 captured in a bakehouse at Poole, Dorset. The specimens of P. americana and P. germanica were caught by hand, and those of the other two species by traps. This probably explains the small number obtained of the young of the two former species, as the larger adults are more easily picked up.

It is possible that some of the young of P. germanica were wrongly included as belonging to P. americana, for both species occur in the same warm houses in the Society's Gardens.

An examination of these specimens showed the percentage frequency of occurrence of individuals with one or more tarsi four-jointed to be as follows:

The examination took no account of the sex of immature individuals, because of the very great difficulty or impossibility of rightly ascertaining it without making a dissection in each case.

In the great majority of cases only one of the six legs bore a four-jointed tarsus, though many individuals possessed the abnormality on more than one leg. This point was examined in detail in rather more than one thousand young and adult individuals of both sexes distributed among three species, with the following result :—

In all four species the posterior pair of legs was the most frequently affected. The following table gives the percentage incidence of four-jointed tarsi among the three pairs of legs :

The abnormal tarsi occurred indifferently on the right and left sides-thus, in 1329 cases in S. orientalis, 661 were on the right and 668 on the left side.

Having set forth the preliminary results obtained, it becomes

necessary to state that at first the frequent occurrence of the abnormal tarsus encouraged a belief that it was a congenital variation. This was supported by several features of the case besides the above. It happened by chance that individuals of P. americana were examined first. Among these there were several cases where one tarsus of a pair was normal, and the other abnormal; and in these cases the tarsi were of approximately equal dimensions and seemed symmetrical till closely examined. Moreover, as shown by Table B, the abnormality occurred with distinctly greater frequency among females than males.

Finally, the abnormal tarsus was found in three individuals of S. orientalis only 4 cm. in body-length, which is the average size of newly-hatched young.

These facts offer material for discussion, but they must be left for awhile, as the evidence they afford in favour of congenital origin of the four-jointed tarsus seems rebutted by that subsequently obtained-that the abnormality is the form assumed by the tarsus when it is a reproduction taking the place of a lost or injured tarsus.

Cockroaches, like other Orthoptera, attain sexual maturity after performing a series of ecdyses, which in some species probably extend over several years. The power of reproducing lost appendages is known to be possessed by members of several of the tribes of Orthoptera. Among the Blattide the reproduction of the antennæ of the Cockroach after amputation was first described by Heineken '.

The evidence given below appears to establish that the fourjointed tarsus arises in connection with the reproduction of the leg when any part thereof has been lost or severely injured during the immaturity of the animal.

(a) Absence of abnormal tarsi in newly-hatched individuals.

A number of egg-cases of S. orientalis were collected and artificially incubated, from which altogether 210 young were hatched. In a few cases limbs were found to have been broken off, but all the tarsi of entire limbs were normal. (b) Relative infrequency of abnormal tarsi in very young

individuals.

That there was a smaller proportion of abnormal tarsi in young than in adult individuals is shown by Table B. This led to an examination of young of different ages, with the following result:

1

"Observations on the Reproduction of the Members in Spiders and

Insects," Zool. Journal, 1829, iv. p. 425.

60*

(c) Evidence from amputation experiments.

In describing these it will be convenient to employ abbreviations, viz:

R and L for right and left.

1, 2, and 3 to denote the respective pairs of legs.

t1, t, and so on for the tarsal joints, starting with the proximal joint.

Altogether the legs of 833 immature individuals of S. orientalis of various ages were mutilated and the animals confined in cages till ecdysis occurred. The mutilations were easily performed under chloroform, but it was found that recovery did not occur unless fresh air was admitted immediately the Cockroaches fell on their backs and ceased to struggle. Of the total number operated on 103 were mutilated in one leg and 730 in two legs. In the tabular summary of these experiments given below, "reproduction" must be taken to imply that regrowth of the parts removed took place, and that the new tarsi were always in a four-jointed condition.

It will thus be seen that about followed by reproduction of the lost appearance of a four-jointed tarsus.

of the mutilations were parts accompanied by the The normal five-jointed

condition never occurred in reproduction. In only two or three

cases were the reproduced parts at all malformed. These will be dealt with later on. The balance of 848 cases in which reproduction was not observed is accounted for by no reproduction occurring at ecdysis or by death before ecdysis. The recent observation by Bordage that the reproduced tarsus of certain Phasmids is invariably four-jointed is of much interest in connection with the above results.

1

(d) Liability to accidental injury in immature individuals.

The evidence already obtained that the occurrence of the four-jointed tarsus in a captured Cockroach indicates accidental injury in the part, was confirmed indirectly by the examination of 1106 young of S. orientalis, when first caught, with the result that 8% were found to be imperfect as regards either the whole or portions of one or more legs. This is a considerably lower percentage than that for the occurrence of four-jointed tarsi recorded in Table B, but the importance of the frequent occurrence of imperfect individuals is increased by bearing in mind that reproduction of lost parts may take place at any of the several ecdyses and that, as has been ascertained by observation, when once a four-jointed tarsus has appeared it is perpetuated through the succeeding ecdysis and almost certainly though all subsequent ones up to their cessation on maturity being attained, which accounts for the higher percentage of such tarsi found in adult individuals.

From another point of view the relation of abnormal tarsi to accidental injury in captured individuals is emphasized by comparing Table D with the following results of examining newly captured S. orientalis for the distribution of imperfect legs :

The long third pair of legs seems to suffer more from their exposed condition as compared with the less extended anterior pairs, and this was observed to be the case with individuals kept in confinement. The tarsi of these, if subsequently reproduced, were invariably four-jointed.

Newport has noticed a similar special liability to injury in the long posterior legs of Scolopendra.

It is of course possible that the four-jointed form of tarsus may be occasionally of congenital origin, but the balance of evidence indicates clearly that in S. orientalis, at all events, it is a result of the loss of the normal tarsus.

1 "Sur la régénération tetramérique du tarse des Phasmides," Compt. Rend. Acad. Sci. séance de 28 Juin, 1897.

2 "On the Reproduction of Lost Parts in Myriapoda and Insecta," Phil.

Trans. 1844.