502

CELL COUNTSBy J. W. STEWART, M.B., B.S., and P. J. CROSLAND-TAYLOR, M.B., B.Chir.

.4ssistant Pathologists, Bland-Sutton Institiute of Pathology, Middlesex Hospital

The invention of the microscope was followedby the discovery of the blood cells. It could beseen that in the case of blood these cells werevery numerous and efforts were made to estimatetheir numbers. In I838 Wagner was able tomeasure the proportion of erythrocytes to leuco-cytes, but it was not until i8;2 that Vierodt wasable to count the absolute number of erythrocytes.His method was to make a known dilution of theblood in a mixing glass and pipette a knownvolume of this dilution on to a slide, in a series ofuniform lines. The slide was dried and all thecells on it enumerated. The method was verytime consuming-three' to four hours being re-quired for a single count. Using this method, hefound that the normal erythrocyte count wasapproximatel, 5 million per c.mm.As leucocytes are outnumbered-approxi-

mately I,OOO times by the erythrocytes it' was notpossible to count sufficient numbers by Vierodt'smethod. In I854 Welcher overcame this diffi-culty by using distilled water as the diluent.This haemolyzed the erythrocytes-leaving onlythe leucocytes to be counted.

In I855 Cramer simplified the whole pro-cedure by inventing a counting chamber. Thiswas a fixed non-separable chamber of knowndepth> and by means of a net ruling in the micro-scope eyepice he was able to calculate the numberof erythrocytes when the magnification of the fieldunder the microscope was known.

In I875 Hayem and Nachet modified Cramer'schamber by making it separable. This they didby putting on a loose coverslip to form one side.

In i877 a type of counting chamber resemblingthose which we know to-day was made by Messrs.Hawkesly and described by Dr. Gowers inLondon.' The main advance was that the netrulings were now on the bottom of the chamberinstead of the ocular of the microscope. Thismade it independent of the magnification of themicroscope with which it was used.

Since I877 modifications to counting chambershave been made by Thoma, Biirker (I9II) andNeubauer as regards ruling and depth, and more

FIG. i -Gower's Chamber showing closed type ofcell. By courtesy of Messrs. Hawkesly Ltd.

recently a bright line chamber has been intro-duced which has a thin coating of metal on thebase of the chamber 'making the ruling morereadily visible.Although counting insufficient numbers of cells

is a source of error, the counting chamber itselfcan be the cause of error. S. Alferow (I884)pointed this out when he measured two chambersmade by Zeiss. One whose depth should havebeen O.I mm. he found on one side to be O.I09mm., on the other o.o88 mm. In the otherchamber whose depth should have been 0.2 mm.was found on one side to be 0.24 mm., on theother O.I96 mm. Some error is unavoidable, butin 1937 a British Standard specification for haemo-cytometers was made and in I953 this wasamended so that those chambers complying toB.S. 748 could not produce an error greater than±4 per cent. and were likely to be less. Thesespecifications could be applied to chambers withrulings of Neubauer, improved Neubauer, Biirkerand Fuchs Rosenthal types.A great deal of valuable work has been per-

formed using these chambers, but they sufferfrom the following disadvantages:

(i) The method is expensive in time and staff.(2) Unless at least 2,000 cells are counted the

accuracy of the method is low.The error in laboratory counts have been

examined by Student (1907), Plum (1936), Berk-son et al. (1940) and Biggs et al. (1948).The main sources of error can be divided into

three:(I) The counting error which was proportional

Protected by copyright.

on June 6, 2021 by guest.http://pm

j.bmj.com

/P

ostgrad Med J: first published as 10.1136/pgm

j.35.407.502 on 1 Septem

ber 1959. Dow

nloaded from

http://pmj.bmj.com/

September I959 STEWART and CROSLAND-TAYLOR: Cell Counts 5036

1) C

C 6 0

4 6

¼!- C

A 6FIG. 2.-Alferow's Chamber, I884, showing open type of cell. By courtesy

of the Royal Society of Medicine

to the square root of the number of cellscounted.

(2) The chamber error due to variations incalibration from one chamber to another.

(3) The pipette and dilution error.Each of these they found in a count of 400 cells

to be 4.6 per cent., making a total coefficient ofvariation of + 8 per cent.The standard method of cell counting to-day

remains in all essentials the same as it was whenGowers introduced his counting chamber over8o years ago, despite the fact that many attemptshave been made to devise more accurate and lesstedious methods. The more important of thesewill now be discussed.

TurbidometryTurbidometry was the first alternative method

to receive attention and was first used to estimatethe percentage of blood cells by Oliver in I896.The method was based on the observation of a

candle flame through a tube of water. To thistube was added blood fixed in Hayem's solution,until the density of the solution extinguished theimage of the flame. The tube was graduated,using normal blood so that the percentage of cellsin anaemia could be estimated.

Ponder (I935), in an experimental approach,found the opacity of cell suspensions dependedon factors other than cell numbers, e.g. the areaof the cells, their shape and refractive index differ-ences between cells and suspension medium. Heconcluded that turbidometric measurements wereinferior to more direct methods of erythrocytecounting. The subject has also been reviewed byWhitlock (I947)-An apparatus depending in part on turbido-

metric principles and originally claiming to be of

value in leucocyte as well as erythrocyte countswas reviewed by Bracket et al. (1953). Theyfound the instrument graduated to give the correctcell count in blood of normal cell diameter butwas dependent on cell diameter outside the limitsof 7.2-8.8pv, thus it gave values 32 per cent. highin a case of macrocytic anaemia and 22 per cent.low in a case of microcytic anaemia. The machinewas found to be unsuitable for leucocyte counts.

In conclusion, opacity measurements are onlyproportional to the erythrocyte count in thosespecimens in which the erythrocytes are similarin size and density, and as this is often in questionthe method has never really been a practicalproposition.

Conductivity MeasurementsThe findings by Stewart in I897 that erythro-

cytes were almost perfect non-conductors ofelectric current compared with plasma or othersolutions of electrolytes, led to attempts to equateconductivity measurements with erythrocyte con-centration (Stewart, I899; Wilson, I905). Thesubject has been reviewed briefly by Hirsch et al.(1950), who, while agreeing with other workersthat there was a close correlation between con-ductivity measurements and packed red cellvolume, showed further a correlation betweenconductivity and red cell count. Further research(Cook, 1952) has shown conclusively that whileconductivity measurements accurately determinered cell mass (or hematocrit), they do not measurethe number of cells which make up this mass.

Electronic Cell CountersAlthough the improvement of pipettes and

counting chambers of to-day allow greateraccuracy in cell counts than those of Berkson

Protected by copyright.

on June 6, 2021 by guest.http://pm

j.bmj.com

/P

ostgrad Med J: first published as 10.1136/pgm

j.35.407.502 on 1 Septem

ber 1959. Dow

nloaded from

http://pmj.bmj.com/

POSTGRADUATE MEDICAL JOURNAL

et al. in I940, the problems of counting sufficientnumber of cells 'remain. The size of the prob-lem is seen when it is realized that io,ooo cellsshould be counted to reduce this error toi per cent.The first approach to the problem was by

Moldavan in I934. He suggested counting ery-throcytes by passing a small volume of-cell sus-pension through a' capillary under a microscopeand installing a photo-cell to operate a countingdevice. No working model was described and itis unlikely with the photo-cells at his disposal thata working apparatus could have been produced.After the development of the photo-multiplier inI940, fresh thoughts were given to the subject.The first electronic machines counted the cellswhich were placed in an unruled countingchamber. One of the following two methodswere employed to scan the counting chamber.

Either the chamber was scanned electronicallyas in the flying spot microscope (Roberts andYoung, 1952), or it was scanned by a mechanicalstage moving it to and fro beneath the objective.The first method is very costly and has not toour knowledge been used in routine laboratories,most of the machines described employ thesecond method.The theory of scanning has been given by

Lagercrantz (I952) and Macfarlane et al. (1959),and successful machines using these principleshave been described by Cooke Yarborough et al.(I954) and Macfarlane et al. (1959).At the present time there are four electronic

cell-counting machines which are commerciallyavailable. The machines are capable of countingleucocytes and erythrocytes. They all include theleucocyte count in the erythrocyte count, butunder normal conditions the leucocyte count is sosmall that it is ignored. If the leucocyte countis very high, it must be deducted from the eryth-rocyte count to give the true reading.

Electronic Counters of the Scanning TypeJarrel Ash Arithmometer

This instrument, measuring 24 in. x I6i in.' xI3j in., is beautifully made and delightfullysimple in operation. We have had the oppor-tunity of testing the machine for a short period.The two counts necessary in chamber scanningmachines are done successively and the true valuefor the cell count is displayed by a pointer on alarge clearly visible dial. The dial incorporatescorrection for coincidence losses so that nofurther calculations are required. One criticismof the apparatus was the difficulty in cleaningand filling the large chamber area so that it wasfree from bubbles. Later we found small scratches

on the base of the chamber giving rise to spuriouscounts. The 'coefficient of variation on a singlechamber filled repeatedly from a stock suspensionof normal erythrocytes between 6 and 7 percent., although the variation on recounting afilled chamber was only 3 per cent. For leucocytecounts the coefficient of variation is somewhathigher, being i2 per -cent. The counting time isapproximately 50 seconds.

In conclusion, this machine is exceptionallyeasy to operate once the technique of filling thecounting chamber is acquired. It is completelyautomatic. The filled counting chamber is placedon the stage and the machine started. There isno focusing or other adjustment to make. Inaccuracy it has little advantage over the visualmethod of counting but is much quicker and lesstiring, so that a larger number of counts can beperformed in a given time.

The Cassella Blood CounterThis is a solidly built instrument, measuring

22 in. x 20 in. x 30 in. It uses a standard micro-scope and the image must be focused on a screen.The counting chamber is of standard design, butunruled and comparatively easy to fill. The twocounts necessary in any scanning machine aredone automatically and subtracted one from theother by the machine itself, the true count beingcalculated from the machine count by means ofa normogram. In experience with an early modelwe found the coefficient of variation, using severalchambers filled from a stock suspension of normalblood, was approximately 5 per cent.As in the arithmometer there was difficulty in

adequately cleaning the chamber before use,and dirt in the unfilled chamber was likely to becounted as cells. For leucocyte counts the co-efficient of variation was ii per cent., which isexplained in part by the smaller number of cellscounted. The counting time was one minute.

In conclusion, we found this machine reliableand easy to use. Once the technique of cleaningthe chamber is acquired the cell count results aredistinctly better than the results of visual countsand, of course, much less tiring to the operator.A great advantage of this machine is that it usesa technique with which laboratory techniciansare familiar, but this is also its weakness becauseunless the image is accurately focused on thescreen the count will be too low. It is possibleto add on extra units to this machine to obtainautomatically data for all size distribution curves.More recently cell-counting machines have been

developed which dispense with the countingchamber and there are at present two such typesavailable.

S-ebtember 1959"S"P

rotected by copyright. on June 6, 2021 by guest.

http://pmj.bm

j.com/

Postgrad M

ed J: first published as 10.1136/pgmj.35.407.502 on 1 S

eptember 1959. D

ownloaded from

http://pmj.bmj.com/

STEWART and CROSLAND-TAYLOR: Cell Counts

The Eel MachineThis counter, manufactured by Evans Electro-

Selenium Ltd., employs the flow principle de-scribed by Crosland-Taylor (I953) to align thecells in single file before counting. The machineand its performance have been described byCrosland-Taylor et al. (1958). It measures 39 in.x 20 in. x 14 in. The method of operation is toplace a suitably diluted sample containing thecells to be counted on the stand in the aperture-raise the operating handle so that some of thefluid is drawn into the machine. Lowering thehandle removes the sample, zeros the electroniccounter and starts the micropump which deliversa known volume of the cell suspension into theflow chamber. The counting time for each countis io seconds. The recommended procedure is totake the mean of four counts of each sample-which increases the counting time to one minute.The count is shown directly on the scale. Thecoefficient of variation is 2.i per cent. for a singleerythrocyte count, or i per cent. for the meanof four counts. For leucocyte counts the coeffi-cient of variation was 4.5 per cent. for singlecounts and approximately 2 per cent. when theaverage of four counts is taken.

This machine is capable of counting rapidlyand accurately and the operating procedure issimplicity itself. However, it is quite a compli-cated machine and the technique of cleaningand lining up the optical pathway does takesome time to acquire. It requires much moremaintenance than the scanner-type machines, butin return offers much quicker and very muchmore accurate counts. Danningen and Grant(1958) have reported that they were able to countplatelets and also to measure the size of the cellscounted using this machine.

The Coulter Electronic Cell CounterThis machine is the latest development in this

field and is manufactured by Coulter Electronics;It was described by Mattern et al. (I957) andevaluated by Brecher et al. (1956). The Coulterelectronic counter uses no counting chamber andis based on the principle that blood cells are poorelectrical conductors compared with saline, andany blood cell present in a small aperture willlower its conductivity. By drawing j ml. of aI: 50,000 dilution of blood through an apertureof Ioop diameter the cells are sufficiently separatedto give individual pulses to an electric currentpassing through the aperture. The time for eachcount is I5 seconds.We have not had the opportunity of testing

this machine ourselves, but have been impressedby its reproducibility at demonstrations. Brecher

et al. (I956) found that the standard error of thismachine was 2 per cent. We have reason tobelieve that the machine has improved since thistest three years ago and that the standard error isnow in the region of i per cent. The machineis easy to use and since it has no optical pathwaythere is little to go wrong and maintenance careshould be minimal. We understand that it willcount leucocytes and platelets if they are pre-pared in suitable suspensions. The only draw-back that we know of in this machine lies in thehigh dilution required-one part of blood in50,000 of diluent. This makes the samples some-what bulky. This machine can also be used toobtain data for cell distribution curves.

ConclusionAccurate cell counts can only be obtained by

using one of the cell counting machines. Theyall count a large number of cells and the best ofthem have a coefficient of variation of approxi-mately i per cent. Whatever principle thesemachines employ all authors agree that themachine counts tie up closely with visual countson normal and abnormal blood specimens.

All four commercially available cell countingmachines are completely satisfactory for erythro-cyte counts. For leucocyte counts there arespecial problems associated with the small num-bers of these cells and also the difficulty inobtaining really satisfactory diluents. Cell debrisand stray particles are likely to be counted asleucocytes and unless the greatest care is taken inseeing that all glassware, pipettes and apparatusis free from them, falsely high counts may beobtained.

Platelet counts can only be done under verycarefully controlled conditions at the presenttime. Here the main difficulty lies in perfectinga reliable technique and until this has been doneplatelet counts will not be possible as a routineon the present machines.No machine at present can perform a differential

count and there is no prospect of any of thepresent machines being able to do so in the future.

SummaryThe development of cell counting has been

traced and alternative methods described anddiscussed. The performance of four commer-cially available electronic cell machines has beenevaluated-and their advantages and disadvan-tages discussed. These machines will counterythrocytes and leucocytes satisfactorily, but notplatelet counts with the techniques and diluentswhich are available at the present time.Bibliography on page 513.

-September 1959 5°5P

rotected by copyright. on June 6, 2021 by guest.

http://pmj.bm

j.com/

Postgrad M

ed J: first published as 10.1136/pgmj.35.407.502 on 1 S

eptember 1959. D

ownloaded from

http://pmj.bmj.com/

September I 959 SHIARP: The As.essment of Platelet Function 513

CREVELD, S., and PAULLSEN, M. M. P. (1952), Ibid., i, 23.CREVELD, S., LIEM KHE HO, and VEDER, H. A. (1958),

Acta haemat. (Basel), I9, I99.DAMESHEK, W., and MILLER, E. B. (1946), Blood, I, 27.DEUTCH, E., JOHNSON, S. A., and SEEGERS, W. H. (i955),

Circulation Res., 3, II0.DONNER, V. I., MACH, O., and BRANCAVA, S. (i958), Acta

haemat. (Basel), 20, 369.EBERTH, J. C., and SCHIMMELBUSCH, C. (i888), Virchows

Arch. path. Anat., 103, 39.FULTON, G. P., ACKERS, R. P., and LUTZ, B. R. (I953),

Blood, 8, 140.GUREVITCH, J., and NELKEN, D. (1956), Ibid., 10, 924.HARDISTY, R. M., and STACEY, R. S. (I955), J. Physiol.,

130, 7II.HARDISTY, R. M., and WOLFF, H. H. (I955), Brit. J. Haemat.,

I1 390.HARTERT, H. (igsx), Z. ges. exp. Med., 117, I89.HAVERBACH, B. J., DUTCHER, T. F., SHARE, P. A., TOMCH,

E. G., TERRY, L. L., and BRODIE, B. B. (I957), New Engl.Y. Med., 256, 343.

HAYEM, G. (1878), C. R. Acad. Sci. U.R.S.S., 86, 58.HAYEM, G. (I882), Ibid., 95, I8.HAYEM, G. (1883), Ibid., 97, 458.HJORT, P., RAPAPORT, S. I., and OWREN, P. A. (I955), Blood,

I0, 1139.HOWELL, W. H. (I9I0), Amer. J. Physiol., 26, 453.IMMERSLUND, 0. (I947), Acta paediat. (Uppsala), 34, 315.INTROZZI, P., and NICOLA, P. DE (I956), Sci. med. ital., 4 656.JACKSON, D. P., HARTMAN, R. C., CONLEY, C. 1. (1953),

Bull. Yohns Hopk. Hosp., 93, 370.JOHNSON, S. G., and SCHNEIDER, C. L. (1953), Science,

117, 229.KOPPEL, U. (I958), Z. Zellforsch., 47, 401.LEEKSMA, C. H. W., and COHEN, J. A. (1956), J. clin. Invest.,

35, 964.LEWIS, J. H., FERGUSON, J. H., SPAUGH, E., FRESH,

J. W., and ZUCKER, M. B. (I957), Blood, 12, 84.LEWIS, J. H., ZUCKER, M. B., and FERGuSON, J. H. (1956),

Ibid., II, 1073.MAcDONALD, L., and EDGILL, M. (I957), Lancet, Ri, 457.MACFARLANE, R. G. (I94I), Quart. J. Med., 34, I.MARX, R., STITCH, W., and EHRHART, H. (I956), Proc. VIth

Int. Cong. Haemat., p. 560.MERSKEY, C. (I950), J. clin. Path., 3, 130.MOOLTEN, S. E., VROMAN, L., and VROMAN, G. M. S.

(I949), Amer. . clin. Path., I9, 814.MORAWITZ, P. (I905), Ergebn. Physiol., 4, 307.MORITA, H. (1958), 'Blood Platelets in Chemical Medicine,'

Tokyo.

MUSTARD, J. F. (I956), Brit. J. Haemat., 2, 17.NICOLA, P. DE (I954), Rev. Hemat., 9, 536.NILSSON, I. M., BLOMBACK, M., JORPES, E., BLOMBACTI.,

B., and JOHANSSON, S. A. (I9S7), Acta med. scand., 159, 179.NOLF, P. (I908), Arch. int. Physiol., 6, I.O'BRIEN, J. R. (i955), Brit. J. Haemat., 1, 223.QUICK, A. J., SHANEBERGE, J. N., and STEFANINI, M.

(I949), Amer. 7. med. Sci., 217, I98.QUIRY, D., CAEN, J., and BERNARD, J. (I958), Rev. Fran;aise,

3, 477.RATNOFF, 0. D., and COLOPY, J. E. (I95s), 7. clin. Invest.,

34, 602.ROSENTHAL, R. L., DRESKIN, 0. H., and ROSENTHAL, N.

(i95s), Blood, 1o, I20.ROSS, S. W., and EBERT, R. V. (x959), J. clin. Invest., 38, I55.SELIGMAN, M. (I957), C. R. Acad. Sci. U.R.S.S., 224, 2192.SEREF-INCEMAN (I955), Sang, 26, I90.SEREF-INCEMAN (x956), Forum Med., 2, 105.SHARP, A. A. (1957a), 'Actes 2nd Symposium Foundation

Baldacci,' Lisbon, p. 72.SHARP, A. A. (1957b), 'Physiological and Antigenic Properties

of Blood Platelets,' p. 76, thesis for degree of M.D., Universityof Edinburgh.

SHARP, A. A. (I958), Brit. 7. Haemat., 4, a8.SOULIER, J. P., and LARRIEU, M. J. (19S4), Rev. Hemat., 9,77.SOULIER, J. P., LARRIEU, M. J., and WARTELLE, 0. (I955),

Acta haemat. (Basel), 14, I6o.SPRAGUE, C. C., HARRINGTON, W. J., LANGE, R. D., and

SHAviLEIGH, J. B. (1952),3. Amer. med. Ass., I50, II93.STEFANINI, M., and MURPHY, I. S. (I956), 7. clin. Invest.,

35, 355-TAIT, J., and BURKE, H. E. (1926), Quart. .7. exp. Physiol.,

x6, 141.ULUTIN, 0. N. (I958), Lancet, ii, S91.ULUTIN, 0. N., and KARACA, M. (I956), Forum Med., 4,257.WEIL-MALHERBE, H., and BONE, A. D. (1954), Nature, Lond.,

174, 557-WEINER, M., ZELTMACHER, K., REICH, C., and SHAPIRO,

S. (1948), Blood, 3, 1275.WRIGHT, H. P. (I94I), . Path. Bact., 53, 255.ZUCKER, M. B. (I947), Amer. Y. Physiol., 148, 275.ZUCKER, H. D. (I949), Blood, 4,631.ZUCKER, M. B., FRIEDMAN, B. K., and RAPPORT, M. M.

(I954), Proc. Soc. e4s. Biol., N.Y., 85, 282.ZUCKER, M. B., and BORRELLI, J. (i9ss), Y. appl. Physiol.,

7, 425.ZUCKER, M. B., and BORRELLI, J. (I956), Amer. . clin. Path.,

26, 13.ZUCKER, M. B., and BORRELLI, J. (I9S9), Y. clin. Invest., 138,

148.

Bibliographv from page 5o; -7. W. Stewirt, M.B., B.S., aid P. 7. Crosland-Taylor, M.B., B.Chir.BIBLIOGRAPHY

ALFEROW, S. (1884), Arch. Physiol. Norm. Path., 3, 269.BERKSON, J., MAGATH, T. B., and HURN, M. (1940), Amer.

3'. Physiol., 128, 309.BIGGS, R., and MACMILLAN, R. L. (1948),J7. clin. Path., x,

269, 288.BRACHETT, R 3., MATTERS, C. F., and OLSON, B. F. (I953),

Amer. .7. clin. Path., 23, 73I.BRECHER, G., SCHNEIDERMAN, M., and WILLIAMS, G. Z.

(1956), Amer. J3. clin. Path., 26, 1439.BURKER, K. (ii9i), Arch. ges. Physiol. (Pflugers Archives), I42,

337.COOK, H. F. (I952), Brit. Y. applied Phys., 3, 249.COOKE YARBOROUGH, E. H., and WHYARD, R. E. (1954),

Brit. J. applied Phys., 5, Suppl. 3, S. 147.CRAMER, A. (I855), quoted from Plum. Nederlandisch Lancet,

3rd Ser., 4, 453.CROSLAND-TAYLOR, P. J. (1953), Nature (Lond.), 171, 37.CROSLAND-TAYLOR, P. J., STEWART, J. W., and HAGGIS,

G. (I958), Blood, 13, 398.DANNINGEN, K., and GRANT, E. H. (I958), Atom Praxis, 4

(Io), Oct. x958.GOWERS, W. R. (I877), Lancet, ii, 797.HAYEM, G., and NACHET, A. (i875), C.R. Acad. Sdi., Paris,

8o, I o83.

HIRSCH, F. G., and TEXTER, E. C. (I950), Blood, 5, 1017, 1036.LAGERCRANTZ, C. (I952), Acta physiol. scand., 26, Sup. 93.MACFARLANE, R. J., PAYNE, A. M. M., POOLE, J. C. F.,

TOMvLINSON, A. H., and WOOLF, H. S. (i959), Brit. J.Haemat., 5, i.

MATTERN, C. F., BRACHETT, F. S., and OLSON, B. J. (I957),Y. applied Physiol., 10, 56.

MOLDAVAN, A. (I934), Science, 8o, I88.OLIVER, G. (I896), Y. Physiol., I9, Proc. Physiol. Soc. XVI.PLUM, P. (1936), Acta med. scand., 90, 342.PONDER, E. (I935), Amer. J3. Physiol., IiI, 99.ROBERTS, F., and YOUNG, J. S. Z. (I952), The Proc. Instn.

elect. Engrs., 99, Pt. IIIA, No. 20.STEWART, G. N. (I897), J. Boston Soc. M. Sc. No. x6, x8, quoted

by Stewart.STEWART, G. N. (I899), J. Physiol., 24, 356.STUDENT (1907), Biometrika, 5, 351.VIERODT, K. (I852), Arch. Physiol. Heilk., II, 26, 327, 854.WAGNER, R. (I838), 'Nachtrage fur vergleichenden Physiologie

des Blutes', Leipzig.WELCHER, H. (I854), Vjachr. Prakt. Heilk. (Prague), 44, xI.WHITLOCK, J. H. (I947), Blood, 2, 463.WILSON, T. M. (I905), Amer. J7. Physiol., 13, 139.

Protected by copyright.

on June 6, 2021 by guest.http://pm

j.bmj.com

/P

ostgrad Med J: first published as 10.1136/pgm

j.35.407.502 on 1 Septem

ber 1959. Dow

nloaded from

http://pmj.bmj.com/