to speak, as labels affixed to certain classes of objects, whereby they become represented by parts of speech and can be referred to briefly.”

H. bigemina has not previously been reported from the Southern Hemisphere, although limited material of haemogregarines obviously close to this species and possibly referable to it has been described from blennies in South Africa (Fantham, 1930) and Fiji (Laird, 1951). During the present studies it was found in the blood of five new hosts:—

Gobiesocidae.

Oliverichtus melobesia (Phillipps)—1'/1, Russell (February 6, 1951); 4/29, Wellington (July 6, September 14, 1951).

Blenniidae.

Ericentrus rubrus (Hutton)—16/20, Wellington (September 17, December 2, 1950; March 15, May 6, July 6, 1951).

Tripterygion varium (Forster)—2/5, Russell (February 6, 1951); 2/6, Auckland (January 12, 1951); 2/9, Tolaga Bay (February 15, 1951); 21/90, Wellington (July 30, August 8, 1949, September 1, October 15, December 26, 1950, March 3, June 6, July 6, September 14 and 29, 1951).

Tripterygion medium Günther—1/36, Russell (February 8, 1951); 6/75, Wellington (July 30, August 7, 1949, October 8, December 25, 1950).

Notoclinus fenestratus (Forster)—1/1, Wellington (October 15, 1950).

The length range of the infected fishes was as follows, the bracketed figures representing the overall length range noted for each species:—

Although a precise study of the seasonal incidence of H. bigemina was not undertaken, the parasite was recorded throughout the year except in April and November, in which months very few fish were collected. The haemogregarine was found in fishes of practically all sizes, the heaviest infections being found as a rule in the younger examples. Its common occurrence in very young fishes of about 2 cm. in length is of considerable interest as regards transmission. Leeches are the vectors of some, if not all, of the haemogregarines of aquatic reptiles and amphibians, but these have not been recorded from any of the New Zealand hosts of H. bigemina. Neither Laveran and Mesnil (1902) nor Neumann (1909) found leeches on the European blennies infected with H. bigeminà, although Abranchus blennii has recently been described from Blennius pholis L. in Wales by Jones (1940). Laveran and Mesnil (1902) recorded leeches (Hemibdella soleae) from soles at Roscoff infected with schizohaemogregarines (H. simondi Lav. and Mes., 1901), but Lebailly (quoted by Neumann, 1909), found only unaltered examples of H. simondi in the gut of the leech Platybdella solea. With the exception of the peritrichous ciliates described herein, the only evidence of ectoparasites discovered in the present studies took the form of some detached legs of an isopod found beneath the operculum of a 45 mm. example of Ericentrus rubrus. Laveran and Mesnil (1902) and Neumann (1909) recorded isopods from European shore fishes, but the latter author could find no trace of haemogregarines in these. At all events, if H. bigemina does

indeed have a blood-sucking invertebrate as its intermediate host, this not only must be of very small size to be able to draw blood—and that without leaving any external sign of its attachment—from fishes only 2 cm. in length, but must also remain attached for only brief periods. Perhaps, as Neumann (1909) suggested, transmission may be effected by way of the intestinal tract, as has since been proved to be the case for some of the haemogregarines of lizards.

The incidence of infection in Ericentrus rubrus was extremely high (80 per cent.). It was 24.5 percent. in Tripterygion varium, 16.7 per cent. in Oliverichtus melobesia, and only 5.9 per cent. in Tripterygion medium.

Laveran and Mesnil (1901) published only a brief account of H. bigemina, their description being confined to intraerythrocytic schizonts and gametocytes. They described and figured the division forms as cylindrical or globose, nuclear enlargement and division followed by cytoplasmic cleavage resulting in the formation of two rounded or pyriform bodies (the rounded ones illustrated in their Fig. 5 being of unequal sizes). These authors stated that the division products develop into “Hémogrégarines adultes” (recognized as gametocytes by Neumann, 1909), elongated and somewhat crescentic bodies swollen anteriorly and tapering posteriorly and attaining 12μ in length by 1.5μ–2μ in breadth.

Merezoite formation in H. bigemina closely resembles this process as described for H. polypartita of the European Gobius paganellus Gmelin by Neumann (1909), with the important difference that it takes place in cells devoid of haemoglobin instead of in red corpuscles as in the latter species.

In the New Zealand material, schizogony leading to the production of merozoites was observed to take place in basophil erythroblasts, small and large lymphocytes, and monocytes. This process was seldom recorded in the blood of the older fishes, these having chronic infections characterized by the presence of very scanty numbers of gametocytes only. Intraleucocytic schizonts and merozoites far out-numbered schizonts and gametocytes in cells containing haemoglobin, in the blood of young fishes having recently-acquired infections.

A 37 mm. example of Ericentrus rubrus collected at Moa Point, Lyall Bay (Wellington) on September 17, 1950, had by far the heaviest infection noted during this investigation. Fifty per cent. of its small lymphocytes and 85 per cent. of its large lymphocytes and monocytes contained schizonts or merozoites, while 4 per cent. of its erythroblasts and erythrocytes held schizonts of the final series or gametocytes. Table III details the results of counts of 100 successive parasitized cells made in the case of small lymphocytes and monocytes respectively, to determine the incidence of the various stages present.

The earliest schizogonies of haemogregarines are those resulting from the initial invasion of various cells of the vertebrate host by sporozoites derived from the invertebrate host. These schizogonies result in the formation of large num-

bers of small merozòites. Some later schizogonies result in the production of smaller numbers of larger merozoites, which invade cells of the circulating blood and there develop into gametocytes.

The earliest stages of H. bigemina encountered were small, ovoid or reniform merozoites occurring free in the plasma of the host and measuring some 3μ by 2μ (Fig. 1). As seen in Giemsa-stained smears, these have a relatively small amount of cytoplasm staining a very pale blue. The relatively large nucleus, which has an irregular outline and measures some 2μ in diameter, is quite rich in chromatin. These merozoites are equivalent to those formed in the earlier schizogonies of typical haemogregarines. The location of the schizonts from which they are derived was not discovered.

Invading various white cells (Figs. 2, 3), the small merozoites develop into schizonts of a pyriform (Fig. 4) or rounded shape.^* At this stage the nucleus of the parasite is pinkish staining and rather diffuse. The cytoplasm is whitish blue and, like that of the equivalent stages of some other sporozoans found in white cells (e.g., the saurian Plasmodium mexicanum Thompson and Huff, 1944), is often so masked by the deeply staining cytoplasm of the host cell that its outline is very difficult to distinguish. The nucleus of the host cell is usually (Fig. 4), though not always, indented by the haemogregarine. Frequently, especially when within the larger leucocytes, the parasite completely penetrates and becomes surrounded by the nucleus of the host cell (Fig. 9).

Binary fission now takes place (Figs. 6, 9), the rounded or oval bodies so formed ranging in size from 2.7μ to 4.8μ by 2.4μ to 4.0μ (av. for 50 examples, 3 7μ, by 3.2μ). Two such schizonts are usually present in parasitized cells (Table III), each of them giving rise to two merozoites, the dominant number of merozoites per cell thus being four. In some small lymphocytes, the bodies resulting from the initial binary fission, instead of rounding up to become schizonts in their turn, develop directly into merozoites. In such cases only two merozoites are formed in the host cell concerned (Table III; Fig. 5).

A further binary fission of one or both of the schizonts of the second series (Fig. 10) may lead to the production of six (Figs. 11, 12) or eight merozoites, or a fourfold division of the initial schizont (Fig. 8) may result in the formation of four schizonts instead of two in the second series (Fig. 13) and consequently of eight merozoites (Fig. 14). Eight merozoites are comparatively seldom formed in a small lymphocyte, although they are quite commonly found in large lymphocytes and monocytes (Table III). Sometimes, but only in the larger host cells, one of the four schizonts formed in the manner related above will undergo a further division. More than five schizonts or 10 merozoites have not been observed in any one host cell.

Merozoites resulting from the intermediate schizogony in the white cells are vermicular bodies, one extremity usually being rounded and the other more or

less pointed. Their dimensions range from 3.9μ to 7.1μ by 0.5μ to 1.0μ (av. for 50 examples, 5.5μ by 0.7μ). The cytoplasm stains light blue with Giemsa, somewhat more deeply so at the periphery, and the centrally (Fig. 12) or posteriorly (Figs. 11, 14) situated nucleus, which occupies the full width and from a third to half the length of the merozoite, is rich in chromatin and stains deep red with lighter red maculations.

These merozoites are morphologically close to those of H. polypartita Neumann, but are relatively more slender than are those of the latter species, which measure 4μ by 1μ (in never-dried haematoxylin preparations). The course of merozoite formation broadly corresponds in the two species, in both of which the commonest number of merozoites formed in a single host cell is four. The maximum number so formed is 16 in H. polypartita as compared with 10 in H. bigemina.

The merozoites are finally freed (Fig. 15), either as a result of their own active movements or of the breakdown of the host cells. Each of them now invades a cell containing haemoglobin, sometimes an erythrocyte (Fig. 20), but much more commonly an erythroblast (Fig. 16). Here they round up to a greater or lesser extent (Figs. 17, 18), usually becoming somewhat pyriform or else cucumber-shaped. Continued growth in size leads to the production of a reniform (Fig. 23), ovoid (Fig. 21) or irregularly rounded schizont. In rare instances two such schizonts are present in one red cell (Fig. 24), in consequence of the invasion of that cell by two merozoites. Schizonts within red cells range in length from 3.6μ to 9.2μ and in breadth from 2.2μ to 4.6μ (av. for 100 examples, 5.5μ by 3.0μ). The nucleus finally divides, the two daughter nuclei migrating to the ends (Fig. 25) or sides (Figs. 19, 24) of the schizont. Cytoplasmic cleavage may take place in the lateral or longitudinal (Figs. 26, 27) plane, and in the latter event may be diagonal (Fig. 28). If cleavage is lateral, the division products are rounded, like those illustrated by Laveran and Mesnil (1901) in their Fig. 5; if it is longitudinal, the young gametocytes are elongate-pyriform in shape (Figs. 29, 30). One red corpuscle which had been invaded by merozoites at two different times was seen. This corpuscle (Fig. 35) contained two developing gametocytes and a schizont as well.

Two gametocytes are invariably formed from each schizont. Although Laveran and Mesnil once observed the commencement of a fourfold division, they never found more than two mature gametocytes within an individual red cell. The nucleus of immature gametocytes is fairly large and of very irregular outline. It is usually situated towards the broader end of the body as in the examples illustrated in Figs. 28–32 and in Fig. 6 of Laveran and Mesnil (1901). The gametocytes are able to move quite actively within the infected erythrocytes from an early stage in their development. They thus adopt a variety of positions with regard to one another and to the nucleus of the host cell. Sometimes they remain in close contact with one another on the same side of the nucleus of the host cell, either with their equivalent extremities corresponding (Figs. 34, 38) or in the tête-bêche position (Figs. 36, 37, 39, 40). They frequently take up positions on either side of the nucleus of the host cell, immature gametocytes often investing this structure very closely (Figs. 30, 31).

With the approach of maturity the gametocytes become longer and relatively more slender, the anterior end being somewhat clubbed and rounded and the posterior one markedly attenuated. The organism is at this stage somewhat

crescentic in shape, but being very plastic it may (as figured by Laveran and Mesnil) be bent into the form of a U or have its fine posterior extremity sharply reflected (Fig. 35). The nucleus migrates into the posterior two-fifths of the body, but does not reach to within more than 2μ of the actual extremity. It becomes an elongated structure, ranging from 2.1μ to 4.4μ in length and from 1.0μ to 1.8μ in breadth. Occupying the full width of the body, it fills from a fifth to a quarter of the total length of the organism. Laveran and Mesnil failed to note any granular inclusions in the cytoplasm of the gametocytes of H. bigemina which they studied. In the New Zealand material, two or three small granules staining blackish red with Giemsa are usually present (Figs. 36, 37, 39), forming a compact group between the nucleus and the posterior extremity of the body. These granules closely resemble those described by Henry (1913c) from H. simondi, a species which Laveran and Mesnil (1901) described but from which they failed to record such granules. According to Henry, the granules of H. simondi are shed by the parasite to initiate a new developmental cycle in the vertebrate host. Similar granules are, however, also present in the intracellular merozoites of piscine haemogregarines (Laird, 1952) (see also Fig. 20). It is probable that they are metabolic products, perhaps associated in some way with the acts of entering and leaving host cells. They may also, like the pigment granules of haemosporidians, be derivatives of ingested haemoglobin. The bulk of the cytoplasm of that part of the body anterior to the nucleus stains a pale whitish blue with Giemsa. It assumes a somewhat darker colour peripherally, and may be more or less macular in appearance. That portion of the cytoplasm posterior to the nucleus stains dark blue. A small polar cap which stains rose-red may be present at the broader end of the body (Fig. 40), but the substance concerned is never as abundant as it is in fish haemogregarines of the rovignensis group (Laird, 1952).

The overall length of mature gametocytes from all the New Zealand hosts ranged from 9.1μ to 14.9μ, and the overall breadth at the widest part of the body from 1.0μ to 2.2μ. The figures derived by Laveran and Mesnil (1901) from their material were 12μ by 1.5μ to 2μ. Average measurements were calculated for 100 mature gametocytes from Oliverichtus melobesia, Ericentrus rubrus and Tripterygion varium, the hosts from which the most abundant material was available. These measurements are detailed in Table IV, together with those of 50 normal erythrocytes from each of the hosts and the percentage increase in length estimated for the 50 parasitized red cells.

As Table IV indicates, there is a striking correlation between the size of the host erythrocytes and that of the mature gametocytes of H. bigemina, the average dimensions of the latter increasing in direct proportion to those of the red cells

of the host. The overall size range of the gametocytes is much the same throughout. Examples of less than 9μ in length are immature, the nucleus still being anteriorly or medianly situated. As regards the upper limit of the size range, only seven of the 100 gametocytes from T. varium and four of those from O. melobesia exceeded 13.9μ in length, the upper limit attained by those from E. rubrus. Both schizonts and gametocytes from all five New Zealand hosts were of identical morphology throughout.

More often than not, the two gametocytes in an individual erythrocyte are dissimilar in size. A count of those in 50 consecutive parasitized erythrocytes chosen at random from among all five hosts revealed that in only six cases were the two gametocytes of exactly the same size.

Erythrocytes containing gametocytes of H. bigemina become somewhat hypertrophied (Table IV), and the nucleus is frequently displaced (Figs. 24, 37, etc.). Hypertrophy may be very marked in doubly parasitized cells (Figs. 24, 35). The staining characteristics of the cytoplasm and nucleus are not affected. The maturation of the gametocyte accompanies that of the erythrocyte itself, as indicated in Table V (compiled from the same preparation from Ericentrus rubrus from which the merozoite counts in Table III were made).

Temporary distortion of the host cell membrane may be brought about by the active movements of the gametocytes within it (Figs. 39, 40). These movements were observed in a fresh cover slip preparation of infected blood from E. rubrus, examined by dark ground illumination. Two mature gametocytes lying on either side of the nucleus of an erythrocyte were seen to straighten themselves out, then contract to the shape of a U. The straightening movement required some 15 seconds for its accomplishment, while contraction occupied only four or five seconds. When the gametocytes, moving synchronously, were both fully extended, the pressure of their extremities against the membrane of the host cell caused this cell to assume an almost rectangular shape, its sides being slightly convex and its ends markedly concave. On the contraction of the haemogregarines, however, the elasticity of the host cell membrane was such as to allow the erythrocyte to resume its normal shape immediately. Other cells were observed in which only one of the two gametocytes was active, and in others again the parasites moved alternately. After about ten minutes of intraerythrocytic activity the gametocytes commence to leave the host cells. Direct pressure against the membrane raises a nipple-like projection (Figs. 39, 40) which finally ruptures, allowing the haemogregarine to glide out into the plasma. Free gametocytes, also those ready to leave the host cells, are bluntly pointed anteriorly (Figs. 40, 43). Like those of other schizohaemogregarines the free gametocytes of H. bigemina move about quite slowly in a gliding manner, always with the broader end foremost. They are often a little narrower and rather

longer (Fig. 43) than intraerythrocytic forms, but otherwise closely resemble these.

In addition to H. bigemina, a number of other species of schizohaemogregarines typically forming two gametocytes in each host cell have been described. Some of these “species” are morphologically indistinguishable from H. bigemina, and their gametocytes fall within the size range of those of the latter parasite (e.g., H. delagei Laveran and Mesnil of European skates). The description of H. gobii Brumpt and Lebailly (1904) is inadequate, and in no way differentiates this haemogregarine of the European Gobius minutus Gmelin from H. bigemina. It is likely, especially in view of the wide range of hosts now known to be parasitized by H. bigemina, that a re-examination of material from the four European species of flat-fishes from which Lebailly (1904, 1905) briefly described four different species of Haemogregarina, will disclose that he was actually dealing with one species only, H. platessae Lebailly, 1904, or perhaps even H. bigemina itself.

Three other species of haemogregarines have been described from blennies. Kohl-Yakimoff and Yakimoff (1915) described H. londoni from the European Blennius trigloides. These authors found only intraerythrocytic schizonts and gametocytes. The schizonts of H. londoni fall within the size range of those of H. bigemina, but the gametocytes do not. These stages measure 9.94μ by 2.48μ to 2.84μ. They are thus broader than the mature gametocytes of H. bigemina, from which they further differ in that their extremities are of the same form and breadth. Sometimes they are surrounded by a cyst-like structure. Such a structure, which may perhaps be formed by the parasitized cell rather than by the parasites themselves, has not been reported for H. bigemina from any of its hosts. Kohl-Yakimoff and Yakimoff noted that some erythrocytes contained only one gametocyte. This could indicate either that the final schizogony does not always take place in H. londoni, or that two gametocytes had developed, one of them already having left the cell. If blood smears are taken from fishes parasitized by H. bigemina an appreciable time after the death of the host, or if cover slip preparations are examined under the microscope for 10 or 15 minutes before permanent smears are made, some of the host cells will be found to have lost one or both of the gametocytes developed within them. In so far as is yet known, H. londoni differs sufficiently from H. bigemina to justify its retention for the present as a valid species.

H. salariasi was described from such limited material from Salarias periophthalmus Val. in Fiji (Laird, 1951) that full and satisfactory comparisons with previously described haemogregarines could not be made. It was hence accorded specific rank purely as a matter of convenience in reference. Only two mature gametocytes were found, these being present in one of the three parasitized red cells located. They were of very small size (7.7μ by 1.0μ and 8.3μ by 1.0μ), but otherwise closely resembled the equivalent stages of H. bigemina, with which H. salariasi may ultimately prove to be conspecific.

Fantham (1930) described H. fragilis from Blennius cornutus (L.) in South Africa. He was dealing with stained material only, and the fact that his preparation contained many free forms indicates that it was probably made some little time after the death of the host. These free forms were of similar size to the free gametocytes of H. bigemina, measuring 11μ to 16μ (mostly about 13μ to 14μ) by about 1.5μ. Intracorpuscular forms measured about 9μ to 12μ

by 1.5μ, thus coming within the size range of the mature gametocytes of H. bigemina. Fantham made no mention of more than one parasite being present within a single host cell, and interpreted the haemogregarines as in the process of entering, not leaving, the erythrocytes. The nucleus of H. fragilis, unlike that of mature gametocytes of H. bigemina, is situated towards the broader end of the body.

Specimen slides of H. bigemina from New Zealand hosts have been deposited in the collection of the Dominion Museum, Wellington (catalogue numbers Z16–17).

Haemogregarina (Hepatozoon?) acanthoclini n.sp. (Pl. 8, Figs. 44–50).

Six of 11 examples of A. quadridactylus collected at Tolaga Bay on February 14, 1951, were very scantily infected with this sporozoan. None of the remaining kelpfish studied were parasitized.

Mature gametocytes were the only stages present, these occupying corpuscles which, despite a superficial resemblance to the smaller leucocytes (Fig. 44), are probably distorted and rather shrunken erythrocytes. The nucleus of the host cell is rather hypertrophied, indented or flattened, and displaced to such an extent that it is squeezed between the parasite and the cell membrane. It is sometimes pyknotic, its chromatin being grouped into a few very deeply staining aggregations.

The parasite is reniform (Figs. 46, 47) or cylindrical in shape, the ends being rounded and of equal width. Only 11 were seen in all. These measure from 8.4μ to 9.4μ by 3.2μ to 4.0μ (av. 8.5μ by 3.5μ). The cytoplasm is dense and homogeneous, and stains a deep and uniform blue, quite distinct from the light blue colour assumed by that of other haemogregarines of fishes. It contains a few small, round vacuoles, and in some cases a dot of chromatic material which stains a bright red colour (Fig. 50). One (Fig. 50), two (Fig. 47) or more (Fig. 48) myonemes may be present. The nucleus is situated at or near the middle of the organism, and is usually round or ovoidal in shape. In one example it extends like a band almost across the body. The nucleus stains a uniform light pink colour. Together with the staining reaction of the cytoplasm and the presence of myonemes, this suggests that the parasite in question may belong to the genus Hepatozoon. Members of this genus have previously been described from reptiles, birds and mammals, but not from fishes. The nucleus ranges from 2.4μ to 3.3μ in its greatest diameter by 1.4μ to 2.6μ (av. 2.9μ by 2.1μ).

In all six infections the parasite rate was less than one per 100,000 erythrocytes.

The haematozoan from A. quadridactylus is quite distinct from any of the haemogregarines previously described from fishes. It is accordingly designated Haemogregarina (Hepatozoon?) acanthoclini n.sp., the generic designation being used in its broadest sense.

A slide designated as the type of this species has been deposited in the collection of the Dominion Museum, Wellington (catalogue number Z18). Paratype slides are in the collection of the Department of Zoology, Victoria University College, and in my own collection.