quite clear that Prof. Erlich fully realised the true nature of this element, yet it is most unfortunate that he should have employed such a term to designate a slightly aberrant type of the finely granular polymorphonuclear cell.

are

All authors are agreed that "pseudo- lymphocytes" generally found associated with definite finely granular polynuclear cells, and are especially abundant in effusions approaching the purulent stage. The nucleus swells up, its outline becomes more nearly regular, and its distinctive character is less marked. The protoplasm fails to stain in the normal manner and the appearance of a mononuclear cell is more or less closely simulated.

It is further stated that portions of the nucleus with a zone of protoplasin are separated from the original cell, and that "the pseudo-lymphote " is produced in this way.

In case twenty-four (pulmonic neoplasm with plural effusion) 21.8 per cent. of polynuclears occurred which were not degenerated, and might have been mistaken for small lymphocytes under a sixth; with an oil immersion lens, however, their true nature became at once apparent. The nucleus remains distinctly "polymorphous," except that its outline is more regular than usual, and the protoplasm differs in no way from that found in the finely granular polymorpho nuclear of the blood. The only striking feature presented by the "pseudo-lymphocyte" is that the amount of this surrounding protoplasm has become reduced to a mere zone.

This atypical variety of polynuclear occurred (slightly degenerated) in two other examples of my series, while Dudgeon and Ross not infrequently found it present in the peritoneal exudate of guinea pigs.

If it were a rule to examine only films prepared from fresh fluid and to always count under an oil immersion lens the "pseudolymphocyte" would hardly ever escape detection or cause any trouble.

Patella adds to the confusion by using the term "pseudolymphocyte" in reference to the small lymphocyte almost constantly found in those primary "idiopathic " pleural exudates which we are about to discuss. He states that these lymphocytes are merely the cast off nuclei of endothelial cells and that he has seen this extrusion taking place. No confirmation of this view is to be obtained. I have occasionally seen an isolated endothelial nucleus, generally with a small tag of protoplasm attached, lying free in the field. Since such a nucleus, when stained by Leishman, in no respect resembles a lymphocyte, it is difficult to understand why this point was ever raised. In counts done on ascitic fluids it is common

enough to see small endothelial cells, but here again with the high powers it is hardly possible to mistake such elements for lymphocytes.

It is also probable that the great variety of staining reagents used by authors (see literature) may account for some of the discrepances. Crude methelene blue, hæmatoxylin and eosin, and Erlich's triacid stain are by no means satisfactory. I believe that if Prof. Leishman's modification of the Romanowsky stain were exclusively used for this work, our cell counts would gain in accuracy.

Passing on to some technical points it will be found that Widal in his paper states that from 20 to 40 c.c. of an effusion must be injected into the peritoneal cavity of a guinea pig, if positive results are to be insured. Lovell Gulland puts the limit at 70 c.c. and has waited 13 weeks for a result.

It is certainly necessary to deal with large quantities of fluid owing to the scarcity of the tubercle bacillus in these examples, but nevertheless in practice this method has serious drawbacks. The animal becomes almost at once acutely ill and may die within a few hours. On examination no evidence of peritonitis can be found, but frequently a large quantity of unabsorbed fluid is present. It is probable that these ill effects are purely mechanical and they have been successfully avoided in this work by using the sediment from 100 c.c. of effusion centrifuged at high speed.

A reference to Table I. will show that this plan has been attended with very constant results, and it is unnecessary to inject more than 10 c.c. Since using this method I find that Prof. Osler also refers to the advantages of centrifugation as a preliminary measure, but apparently this course has been adopted in only very few instances.

The importance of a complete autopsy in these animal experiments cannot be over-estimated. One example may be cited in which evidence of infection rested solely on the finding of four or five tubercle bacilli in a smear preparation from one of the retroperitoneal glands.*

In many instances the microscope will establish the presence of early lesions which must inevitably have escaped the most careful macroscopic scrutiny.

Table I. gives the results obtained from seventeen examples in which the small lymphocyte was the principal cell. In those cases classed as "not proved" it was either impossible to perform an inoculation experiment or the guinea pigs died at an early stage. It will be observed that in fourteen examples the percentage of small lymphocytes is very high and that the presence of other varieties of white corpuscles is fully accounted for by blood contamination. In such primary tuberculous cases, confirmed by inoculation, tuberculin injection, or autopsy, Widal and Ravaut found lymphocytosis almost exclusively. Finely granular polynueclear cells, if present, are often less than ten per cent.

Positive evidence of tuberculosis was obtained in eleven out of twelve of my cases which were tested by inoculation experiments or otherwise, and this amounts to 91.6 per cent.

If the individual examples in this table be examined the following points are of interest.

In case 2 (49.8 per cent. of small lymphocytes) the cytological count did not appear to point to tuberculosis; since, however, no animal was inoculated this instance must be taken as wanting proof.

Mr. L. S. Dudgeon has recently had a precisely similar experience,

There were in addition to the S.L. 32-4 per cent. of endothelial cells present showing mitotic figures. The patient made a complete recovery and reported herself quite well eleven months later. Widel quotes a somewhat similar case of sero-fibrinous effusion complicated with signs of softening at one lung apex. He found numerous small lymphocytes and placards of endothelial cells. Two guinea pigs inoculated with 20 and 40 c.c. of the fluid proved negative.

Case 4 has already been referred to; it shows the great importance of searching for tubercle bacilli in smears made from the retroperitoneal, iliac, and lumbar glands and from the spleen. before the guinea pig is passed as not infected.

In case 11 with 890 per cent. of small lymphocytes the most careful and complete examination of the animal failed to show any evidence of tuberculosis. I prefer not to attempt to explain this result away by postulating an error of technique, although it would. be a fair contention.

Case 31 is of special importance as illustrating what has already been stated as regards the rapid disintegration of the cells in these fluids. Here thirty-six hours had to elapse before the fluid could be examined and by mistake it was centrifuged in the routine manner. Only a granular detritus was obtained. Two samples of this exudate were injected into the peritoneal cavities of a pair of guinea pigs in order to gauge the relative toxicity of the upper as opposed to the lower layers of 100 c.c., which had been centrifuged at high speed to obtain the micro-organisms. The animal inoculated with the sediment developed obvious macroscopic tuberculosis in 56 days, while, as far as could be ascertained by naked eye examination, the other was not diseased. Nevertheless under the microscope early tuberculosis lesions were detected in the spleen, and tubercle bacilli were found in smear preparations made from the retroperitoneal glands. The value of centrifugation. as a preliminary to intraperitoneal inoculation is thus well shown. A few degenerated lymphocytes were seen in films prepared direct from the pleural fluid, but in the table cells have been returned as absent. One or two other examples of this type occurred, but they were rejected as unsuitable for further experiment because the cells were not in good condition.

According to Widal the normal cerebrospinal fluid contains two or three small lymphocytes per fieid of the oil immersion lens, but Nageotte, Babinski, and Jamet, all comment on the great scarcity of cells in the normal state. It is not easy to dispose of this difficulty, for any opportunity of examining the human cerebrospinal fluid. apart from disease rarely occurs in Great Britain,

Only two cases of tuberculous meningitis (proved at autopsy) are included in Table I., and the cytology of the cerebrospinal fluid in both differed in certain respects from the formula of Widal and Ravaut.

These authors, as the result of numerous careful observations and experiments, find that in tuberculous meningitis relatively large numbers of polynuclear cells frequently occur, but that even in blood stained specimens of fluid the small lymphocytes are always in excess. They further urge that a comparatively high percentage of polymorphonuclears mixed with the lymphocytes does not invalidate the diagnostic importance of the mononuclear elements, although a differential count may be needed to establish the excess of the latter cells. In my cases the lymphocytes predominated, but were not numerous, and polymorphonuclear cells occurred but rarely.

In neither instance was the full differential count of 500 cells possible, and the small lymphocytes amounted to 93 and 77 per cent. of the total cells present. Bernard found in one tuberculous case 140 lymphocytes (82-4 per cent.), and 30 polynuclear cells; four days later there were 36 lymphocytes and 157 polynuclears. The increase in the polynuclears coincided with a secondary infection by pyogenic organisms, and this may be the explanation of many of these cases. I may add that Mr. L. S. Dudgeon has noted a very marked predominance of lymphocytes in a number of cases examined. It will therefore be seen that the findings of many different observers do not absolutely agree with those of Widal in this respect.

TABLE II.-The peritoneal fluid in the inoculation experiments.