the posterior three dorsal teeth and regeneration has begun. Fig. 16 is a case where a portion of the rostrum has been broken and some of the dorsal teeth have been damaged. A slight malformation in the regenerating portion can be seen. Figs. 19 and 21 are examples where the broken rostrum has begun to regenerate a new anterior portion and teeth, though these are still smaller and more crowded than on a normal rostrum.

Only a few cases showing original damage were observed. This damage could be divided into two different classes—i.e., those with loss of the anterior portion of the rostrum including the 1st dorsal tooth, and those with loss of almost the entire rostrum. Only damage where the scars had become healed over with a chitinous.

Text-fig. 3.—Fig. 1—Diagram of rostrum and dorsal surface of carapace, to show measurements taken. Fig. 2—Outline of rostrum, before ecdysis, of specimen from Pt. Howard, Wellington. Fig. 3—Enlargement of anterior portion of Fig. 2, showing new rostrum under old. Fig. 4—Portion of ventral surface of rostrum of a ♀ specimen from Pt. Howard, showing split ventral tooth. All specimens are Palaemon affinis and setae are omitted.

membrane was included, thus excluding those specimens which had been damaged during collection or after death. The numbers can be summarised as follows:—

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Therefore 5% of all specimens examined showed damaged and unregenerated rostra, or in the case of some showing loss of the entire rostrum, not more than one moult had occurred. Those showing loss of the anterior portion of the rostrum including the 1st dorsal tooth were most common, 2.3% of all specimens, while 1.7% showed loss of the entire rostrum and 1% had passed through one moult since losing the entire rostrum. Therefore the combined incidence of loss of the entire rostrum (2.7%) was slightly higher than the incidence of loss of the anterior portion only. Thus damage to the rostrum is commoner than has been previously recognised. The Island Bay samples showed little original damage, for example sample 2 had only 1 specimen showing loss of anterior portion and none with loss of the entire rostrum, while a collection taken in the same pool as sample 2, about 8 months later, showed no original damage at all. The sample from the rock pools of Kaingaroa, Chatham Islands, had the greatest percentage of damaged specimens (15%) while the collection from the sandy, protected lagoon at the Chathams showed no damage at all. The percentage of damage, however, in the Kaingaroa sample, is probably abnormally high due to the small size of the sample.

Different stages in the regeneration of the rostrum give rise to a large number of the variations in the position of the most anterior dorsal tooth and in the number of ventral teeth. That is to say, these are not stable features, during the next few moults they may be expected to develop closer to the genetic formula. If in Fig. 1 (Text-fig. 2) the tip was broken off, the “quoianus” type, as in Fig. 4, would be an intermediate step in regeneration. If in Fig. 2 the tip, including the 1st ventral tooth was broken off, then Fig. 12 would be one of the early stages in regeneration. That this has, in fact, happened in the case of the specimen from which Fig. 12 was drawn is the more probable because of its singular position in the scatter diagram (Text-fig. 4) and by the fact that its rostrum is much shorter than its carapace (0.82:1). Figs. 2 or 6 are from specimens which on losing their tips would give a rostrum as in Fig. 5 at one stage in regeneration. Many more examples could be given and all other types could be derived from specimens such as Figs. 1, 2, 6 and 7, or specimens of the same type but with different relative positions of the dorsal and ventral teeth.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Text-fig. 3, Figs. 2 and 3, from the one specimen, gives an example of restorative regeneration taking place at a moult. The integument about to be lost at ecdysis has the formula of 9/4, both the 1st dorsal and the 1st ventral teeth are very small in size and the anterior portion of the rostrum is crowded and small. But in Fig. 3 the new integument showing within the old, which has pulled away slightly from it as one of the first signs of an impending moult, has large 1st dorsal and ventral teeth.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

and appears to have a new dorsal tooth (T), anterior to the old 1st, which would give the new rostrum a formula of 10/4 and an almost bifid tip.

This demonstrates that in regeneration the dorsal and ventral rostral teeth are reformed posteriorly first so that the 1st dorsal of the above specimen would become the 2nd dorsal after ecdysis. In fact the teeth taken as 1st dorsals in the specimens of groups C and D of the diagram (Text-fig. 4) including the “quoianus” types, are only temporarily 1st dorsals, and after one or more ecdyses would no longer be the most anterior teeth on the dorsal surface of the rostra.

In Text-fig. 4 one can see clearly one major grouping which contains the specimens with undamaged or completely regenerated rostra, and two minor, which contain individuals having the rostra in various incomplete stages of regeneration. The major group includes both groups A and B, and forms 62% of the 50 specimens taken as the sample. The two minor groups are smaller, the first, consisting of group C and including the “quoianus” types, forms 24%, and the second, group D and the individual with two ventral teeth, forms 14%. In the full sample of 55 specimens there was only one specimen with the anterior portion of the rostrum broken off and showing no regeneration. As the specimens forming groups C and D could be regenerating from damage of any degree, it is difficult to gauge from these figures the percentage of damage to be expected in a sample or the number of moults needed to complete the regeneration after any type of damage. It is indicated in the diagram that it takes at least two moults to regenerate after loss of only the.

Text-fig. 5.—Fig. 1—histogram of the rostral length expressed as parts per 100 of the carapace length for Palaemon affinis. Single line—50 specimens from Island Bay (sample 2). Double line—16 post-larval specimens from Island Bay, Wellington. Fig. 2—Relationship between the number of dorsal and the number of ventral teeth on the rostrum of 16 post-larval specimens of Palaemon affinis from Island Bay, Wellington.