No. 4111.

JUNE 14, 1902.

Presidential AddressON

THE ANNUAL VARIATION OF PUERPERALFEVER COMPARED WITH THAT OF

SOME ALLIED DISEASES.Delivered at the Opening of the Session of the Hunterian

Society

BY A. L. G A L A B I N, M.D. CANTAB.,F.R.C.P. LOND.,

OBSTETRIC PHYSICIAN TO GUY’S HOSPITAL.

GENTLEMEN,-In selecting a subject for a short addresson the commencement of my duties I have chosen one

which is connected with my own speciality but is related

also to general medicine and surgery-namely, the varia-tion in prevalence of puerperal septicaemia in different

years as compared with that of certain allied diseases-

erysipelas, general septicaemia and pysemia, scarlet fever,and acute rheumatism. The method I have followed is nota novel one but is in the main that adopted by Dr. G. B.Longstaff in his work, " studies in Statistics," published in1891. In this work is a table, which I have reproducedin part in my "Manual of Midwifery," showing, by thegraphic method of curves, the variations of the above-mentioned diseases, as well as several others, for the25 years 1855-1880 in England and Wales. I now pre-sent to you two diagrams (Figs. 1 and 2) showing thevariations of these five diseases. One refers to Englandand Wales and includes 19 years-1881-1899-commencingafter the end of those years tabulated by Dr. Longstaff.The other refers to London and includes the 20 years 1881-1900. The figures are taken from the reports of theRegistrar-General, but these, unfortunately, are published Ionly a long while after the date referred to. Thus the annualreport for 1900 is not yet published and I cannot thereforecarry the statistics for England and Wales further than theend of 1899. By using, however, the short annual summaryof weekly reports I am enabled to carry the statistics forLondon to the end of 1900 and so complete the 20 years.These annual reports of the Registrar-General have thus onlya retrospective interest in any bearing they have uponmedicine or surgery, for a warning nearly two years afterdate of any existing or threatened epidemic prevalence ofdisease is too late to be of service. In the weekly reports toomany diseases are grouped together to furnish informationas to the prevalence of any one except a very few zymoticdiseases. It appears a great pity that there cannot be givenat least in the quarterly reports as much information as tothe prevalence of particular diseases both in England andWales and in London as is now to be found for London onlyin the short annual summaries of weekly reports.The following is the plan which I have adopted in con-

structing the diagrams. The curves by their ordinates repre-sent the percentage above or below the mean of the

mortality of the disease in question for any given year.This is the method of representation followed by Dr. Long-staff, and before him by Buchan and Mitchell in the .7M’?’M<xZof tlte Scottish Meteorological Society for 1874. It avoids the

difficulty which would otherwise exist in comparing thecurve of a disease having a high mortality with that of onehaving a comparatively very low one and allows the rela-tive variations of each to be readily seen. I have not, how-ever, taken the figures from the mortality of the differentdiseases per 1,000,000 living, as given in the Registrar-General’s reports, according to the estimated population. Ihave taken the actual number of deaths for each year andfirst reduced each of these to the percentage above or belowthe mean and embodied them in curves. The mean, as

corrected for population, is there represented as a slightlyascending diagonal line for each of the 10 years 1881-1890,1891-1900, according to the results of the censuses in 1881,1891, and 1901. It is then easy to see in this preliminarydiagram the percentage of mortality in each year above orbelow the corrected mean. Thus a second diagram is con-structed in which the mean is reduced to a horizontal line, as

in the figures before you. This method ought to be some-what the more accurate, since of late years each census hasshown that the estimated population has been put too highboth for England and Wales and for London.The uppermost curve on each diagram represents the

annual rainfall in inches at Greenwich. This is an invertedcurve, the ordinates being drawn from above downward, sothat each elevation of the curve represents a minimum andeach depression a maximum rainfall. The reason for invert-

ing the curve in this manner is that it is found that, on thewhole, a minimum rainfall generally corresponds to a

maximum mortality from the diseases in question, and aminimum rainfall to a maximum mortality, a result whichmay at first sight appear surprising as regards acute rheuma-tism. It must be admitted, however, that this relation is byno means clear or decisive in the 20 years under considera-tion. A dry year in 1884 is associated with a slightmaximum in two only of the curves-that of erysipelas andthat of septicsemia. A dry year in 1887 is associated witha maximum in three curves-puerperal fever, erysipelas,and scarlet fever; a less extremely dry year in 1890with a slight maximum in four curves-puerperal fever,erysipelas, scarlet fever, and rheumatic fever. Themost marked association is that of the dry year 1893with an isolated maximum which is the most prominentfeature of four out of the five curves throughout the whole20 years, that of septicaemia being excepted, and which ismere developed in London where one would expect theassociation with rainfall at Greenwich to be closer than inEngland and Wales. In observing more particularly thepeculiarities of 1893 it is notable that it followed anotheryear, 1892, which was dry though less extremely so ; andthat in one of its quarters, April to June, there occurred theminimum quarterly rainfall of the whole 20 years-namely,1’47 inches. There was also a second period of drought,less extreme, for eight weeks in August and September and alarge proportion of the whole rain of the year fell in sixweeks, the eighth, ninth, twenty-eighth, forty-first, forty-second, and forty-sixth weeks of the year, only one of theselying between the beginning of March and the end ofSeptember.

’

Thus far there appears a distinct association of dry seasonswith maximum mortalities of the five diseases. But since1893 there have been two dry years, 1895 and 1898, occur-ring singly, and not associated with any maxima of thesediseases but rather with minima. But in these years therain has been more distributed. Thus while in 1893 therewere only 27 days on which rain fell in the four monthsMarch to June, in 1895 there were 45 and in 1898 57. Inthe diagram constructed by Dr. LongstafE for the preceding25 years, 1855-1880, there appears a much closer corre-

spondence between maximum mortalities of these diseasesand minima of a curve constructed by adding together thepercentages above or below the mean of the amount of rain-fall and the number of days in the year on which rain fell.The conclusion is that there is a connexion between unusual

dryness and excessive mortality of these diseases, but thatthis connexion has not been apparent since 1894, havingprobably been overbalanced by other causes of improvedmortality.The meteorological character of the seasons might increase

the prevalence of disease either by promoting the growth ofpathogenic microbes outside the body or by facilitatingtheir diffusion or conveyance. It appears inconceivable thata dry season can, while it lasts, promote the growth ofmicrobes outside the body, or even help to preserve themalive, though it can be readily understood that damp onesmight do so. If very dry periods therefore promote thespread of microbic diseases the probable conclusion is thatthey do so by aiding the dissemination of microbes in theform of dust. An example may be cited tending to show theoccasional importance of dust conveyance of contagion evenin the case of enteric fever, which is generally water-borne-namely, the recently reported case of Bermuda. In Bermuda,enteric fever had long been very prevalent, so long as thedry earth system of latrines was in use. On the change ofthis system to that of water-borne sewage the prevalence ofthe disease was greatly diminished. It is obvious that thewater-borne sewage would increase, rather than diminish, therisk of contamination of water, while the dry earth systemmight promote the conveyance of contagion by dust.

In order to test whether the effect of drought on thespread of disease is immediate and direct I have constructeda diagram (Fig. 3.) showing the weekly variations of the

AA

1672 DR. A. L. GALABIN: THE ANNUAL VARIATION OF PUERPERAL FEVER. [

1673

five diseases in London for the remarkable year 1893, inwhich four of them showed a high maximum of mortality.The numbers are reduced to percentages above or belowthe mean ; and to render the curves more continuous Ihave adopted what is known as Bloxam’s method. Insteadof taking the mortality of each single week as the ordinatesof the curve I have taken the mean mortality of eachconsecutive three weeks. This method tends to elimi-nate casual variations when the number of cases underconsideration is not great enough to eliminate them.The continuous lines show the weekly variation of thefive diseases in 1893 ; the dotted lines the mean weeklycurves calculated from the 10 years 1883-1892. Itwill be seen that the scarlet fever mortality is abovethe mean of the season throughout the whole year. Thechief excess occurs through the second half of the very dryperiod and two months following, a great part of these

being also a period of drought. But, on the contrary, theexcessive mortality of puerperal fever, erysipelas, and rheu-matic fever does not occur in the dry period but forms anexcess of two normal rises-one in August and the other atthe end of October. Each of these culminates from two tofour weeks after a heavy rainfall. It would seem, therefore,that in scarlet fever drought may directly promote contagionby the spread of dust, but that in septic diseases and rheu-matism the effect only appears after an intervening rainfalland lapse of time. This would look as if it implied agrowth of microbes outside the body following their dis-semination, for the interval is too great to be simply due to

time occupied in the development of the disease, especially-since the greatest maximum of mortality did not occur tillthe end of October.

Comparing the mean death-rates per 1,000,000 living inEngland and Wales for the five diseases, the number for-puerperal fever is 75, for erysipelas 54, for septicaemia andpyaemia 13, for scarlet fever 247, and for rheumatic fever,including rheumatism with affection of heart or pericardium,88. In estimating the mortality of puerperal fever it mustbe remembered that it is limited to one sex and to a com-paratively small number of years in the prime of life. Amongthe relatively large number of deaths from scarlet fever agreat proportion occurs in children. It is a notable factthat the mortality of puerperal septicaemia is nearly sixtimes as great as that of all other forms of septicaemia andpyaemia together. For London the mean death-rate ofpuerperal fever is 9’3 per cent. lower than for England andWales ; of erysipelas, 14-7 per cent. higher ; of septicaemia.and pysemia, 46 ’1 per cent. higher; of scarlet fever, 0-8per cent. lower ; and of acute rheumatism, identical. Thespecially high excess in London for septicaemia and pyaemiais probably eonnected with the large proportion of patientswho come to London for treatment in hospitals and otherinstitutions. It cannot be taken as showing any excess ofseptic microbes in London, much less any inefficiency inantisepsis on the part of London surgeons. The lower ratefor puerperal fever in London, as contrasted with the higherrate for the other septic diseases, is satisfactoryfor obstetricpractice in London. The death-rate in England and Wales

P. F., Puerperal fever.B., Erysipelas. Sep., Septicaemia and pyaemia. Sc., Scarlet fever. Rh., Rheumatic fever with rheumatismof the heart. Ra., Rainfall at Greenwich. The continuous lines denote the weekly mortality in 1893. The dotted lines denotethe mean weekly mortality of the ten years 1883-1892. The numbers have been adjusted by Bloxam’s method for the eonstructionof the curves, and reduced to percentages above and below the mean.

1674

appears to be raised by that of remote country districtswhere medical aid is difficult to obtain and uninstructedmidwives are more employed.The curve of mortality for puerperal fever (Figs. 1 and 2)

shows a decided downward tendency, like that of theother four diseases, and this is the more satisfactory sinceno decided tendency of the kind was shown for any of themin Dr. Longstaff’s diagram for the preceding 25 years. It

begins at 16 per cent. above the mean for England andWales and 40 per cent. for London. It finishes at 22 per cent.below the mean for England and Wales and 50 per cent. forLondon, the lowest point reached. This last result, how-ever, was attained in 1900, for which year the figures forEngland and Wales are not yet available. Since 1888 it hasnever been above the mean except for the remarkable risebeginning in 1892 and culminating in 1893, which is more orless common to all the curves but especially marked in

puerperal fever, in erysipelas, and, in London, in scarletfever. Its relation to the exceptionally dry early summerquarter has already been noted. In puerperal fever the riseis more marked in London, where it reached 44 per cent.above the mean, as compared with 34 per cent. for Englandand Wales.

Since it was formerly a controverted question whetherpuerperal septicaemia is ever epidemic it is of interest tonote that its degree of variation is fairly comparable withthat of the other four diseases. It is considerably lessthan that of scarlet fever, reaching only 44 per cent. abovethe mean in London, as compared with 108 per cent. for iscarlet fever. It is somewhat less than that of erysipelas iand septicaemia but decidedly greater than that cf rheumaticfever, which for England and Wales does not rise more than20 per cent. above the mean, though in London in thenotable year 1893 it rose 38 per cent. above the mean. Themean death-rate for puerperal fever is greater than in thepreceding 25 years-75 as compared with 63 per millionliving, but this appears to depend, partly at any rate, upona change in the practice of the Registrar-General. Since1881 it has been the custom, when any woman’s death isreturned as due to septicaemia or peritonitis, to send a letterto the certifying practitioner inquiring whether there hasbeen recent parturition. In the first year of this practice14 per cent. of deaths were added in consequence to the

heading of puerperal fever. The numbers, therefore, are

strictly comparable throughout the years recorded in mytables but not with previous years.As regards the curve of erysipelas its most marked feature

is its close resemblance, almost amounting to identity, tothe curve of puerperal fever. Almost every deviation inthe one is represented in the other, except that a small

secondary maximum, which in erysipelas occurred in 1882,is deferred to 1883 for puerperal fever, both in Londonand in England and Wales. Further, in those respectsin which the curve of erysipelas for London differs fromthe curve for England and Wales, it is followed by thecurve of puerperal fever for London, so that the latter

corresponds much more closely to the curve of erysipelas forLondon than to the curve of puerperal fever for England andWales. I would note especially the extra height of the

great maximum in 1893 and the pointed instead of squareshape of the rise in 1887-88. On the other hand, whenthe curve of puerperal fever is compared with that of

septicaemia and pyaemia, although there is a generalresemblance and most of the elevations and depressionscorrespond, there is a less close resemblance between therelative height of the maxima, as may be noted especiallyin those from 1888 to 1893. In these respects the curves forsepticaemia and pyaemia for England and Wales agreetogether and differ from the corresponding curves forpuerperal fever. The chief difference between the curves oferysipelas and puerperal fever is that the general downwardtendency is less in the latter for England and Wales. InLondon this is only the case up to the year 1886 ; andafterwards up to the last year recorded the improve-ment is slightly greater in puerperal fever. A stillmore remarkable proof of the close relation of erysipelasto puerperal fever will be found in comparing thecurves of weekly variations shown in Fig. 3. Not only Iis the mean curve for 10 years practically identical for .1the two diseases ; but the same is true of the very com-plicated curve showing the weekly variations in 1893. ’,Every cusp and turn and every date when the curves ’,cross the mean correspond. Even the amplitude is similar,the only difference being that erysipelas varies a very little ’i

more widely both in the mean curve and that for 1893.This identity becomes even more striking if the curves are

compared with the totally different curve of septicaemia andpyaemia. This resemblance of the curves confirms theevidence of bacteriology as to the close connexion betweenpuerperal fever and erysipelas. I think, indeed, that theinference may be carried somewhat further and that it may’be concluded that a variety of streptococcus closely alliedto that of erysipelas is more generally concerned in fatal

puerperal fever than in other forms of septicaemia and

pyaemia, and that the contagion of erysipelas is more

dangerous to the puerperal woman than that of any otherseptic disease other than puerperal fever itself. Associatedwith the general descent of the curve for erysipelas is thefact that the mean death-rate per 1, 000, 000 living is only 54for England and Wales, as compared with 92 for the

previous 25 years, during which no decided downwardcourse was manifest. The curve begins for England andWales at 55 per cent. above the mean and ends at 52 percent. below it.The curve for septicaemia and pyaemia shows a general

descent rather less than that of erysipelas, beginning at65 per cent. above the mean and ending at 24 per cent. belowit for England and Wales. In London, however, there is aremarkable rise for 1900 to 35 per cent. above the mean, forwhich I can offer no explanation. The corresponding figurefor England and Wales is not yet available ; but the curveshows the moderate rise for 1899 which in London,preceded the much greater one of 1900. The mean weeklycurve of septicaemia and pysemia (Fig. 3) has some notable. differences from those of puerperal fever and erysipelas.All have maxima in early November and again in January,but in septicaemia and pyaemia the earlier maximum is thelarger, in the other two the latter. Again, septicaemia andpyaemia have a high maximum in April which in the othertwo is represented only by a slight double elevation. Theweekly curve for 1893 shows wide oscillations which maybe casual variations due to the absolute number of casesbeing small.The curve for scarlet fever (Figs. 1 and 2) shows the

most marked descent of all, beginning at 106 per cent. abovethe mean and ending at 60 per cent. below it. The meandeath-rate per million is only 247 for England and Wales, ascompared with 854 in the 25 years preceding; and from1885 onward it is only about 168. The main drop tookplace in the years 1882-85 and was doubtless due to theisolation of patients in fever hospitals. Up to 1891 there islittle resemblance of the curve to the three preceding forEngland and Wales, but more after that date, all havingmaxima in 1893 and three of them in 1896, three of themminima in 1895 and all of them minima in 1898. In Londonthe curve for scarlet fever has a somewhat greater resemblanceto that of erysipelas and puerperal fever from 1886 onward.The more marked epidemic rises in scarlatinal mortality inthe preceding 25 years were not accompanied by seriouscorresponding rises of puerperal fever nor is the drop ofscarlet fever in 1882-1885 associated with any equivalentdrop of puerperal fever. The extra maximum of scarletfever in London in 1887 as compared with England andWales leads to no similar maximum of puerperal fever, butthe curve of that disease has a rise only corresponding tothat of erysipelas. The mean weekly curves of puerperalfever and scarlet fever are quite different (Fig. 3), scarletfever rising slowly to a maximum at the end of October andfalling considerably before puerperal fever rises to its mainmaximum in early January. The diagrams therefore give nosupport to the view formerly held extensively in England thatcontagion from scarlet fever is an important cause of puerperalsepticaemia. Such similarity as exists in the curves maybe explained partly from a similar seasonal influence ondifferent microbes, as by a period of unusual dryness.Again, all would admit that scarlatinal throats may be com-plicated by streptococci ; and that thus a septic infectionmay be carried from a scarlatinal patient to a puerperalwoman. There is a further question whether a year in whichstreptococci are unusually abundant or easily diffused maynot increase the mortality of scarlet fever by causing septiccomplications ; and the same consideration would apply alsoto rheumatic fever. I have no evidence to offer on this

point, but some light might be thrown upon it, in the caseof scarlet fever, by comparing the mortality with the numberof cases notified. The latter has only very recently beenincluded in the returns.

I have included the curve representing rheumatic fever,

1675

including rheumatism with affection of the heart or peri.cardium (but not chronic valvular disease), as a disease allie&egrave;to puerperal fever, because in the diagram of Dr. Longstafifor the preceding 25 years rheumatism of the heart’shows a curve closely resembling that of puerperal fever ancerysipelas except that it is steadily ascending. Since1881 rheumatism with affection o’f the heart or peri.cardium has been massed together with rheumatic feveiin the Registrar-General’s returns and the result isshown in my diagrams. The resemblance to the curves

of erysipelas and puerperal fever is much less thanunder the former arrangement. Up to 1891 it is scarcelyvisible, as in 1887-88 a depression corresponds to the risein the other curves for London, but from 1892 onward thereare three maxima and two minima correspondent, thoughonly one of them, the maximum of 1893, is well marked inall. The mean death-rate per million living is 88, as com-pared with 46 in the previous 25 years for rheumatism of theheart only. The diminution of similarity produced by theaddition of rheumatic fever seems an argument in favour ofthe view that fatal heart affection in acute rheumatism isoften due to streptococcic complications and that thus rheu-matism of the heart is more closely allied to septic affectionsthan rheumatic fever in general. The curve shows less withvariation than the others, but its general tendency is down-ward, though the improvement is less marked than in theother diseases. What improvement has occurred must beascribed to general hygienic measures or improved treatment,since precautions against sepsis or infection can scarcely haveany scope here. The mean weekly curve of rheumatic fever(Fig. 3) has some resemblance to that of erysipelas or puer-peral fever. Both have maxima in early November andagain in early January, but in rheumatism the former issomewhat greater instead of the latter. Both show slightrises in April and in late August and have their minima nearmidsummer.

In conclusion, I think that among the practical inferencesto be drawn from the diagrams are the following : 1. Thatwhile a satisfactory improvement in the mortality of

puerperal fever has occurred for the first time within thelast 20 years, a further development of antiseptic midwiferyis still required in England and Wales outside London tomake the improvement equal to that already attained inother septic diseases. 2. That the relation of fatal puerperalfever to erysipelas is still closer than has been generallysupposed, and much closer than that of either disease tosepticaemia and pyasmia generally. A consequence of thisseems to be that an anti-streptococcic serum derived fromerysipelas would be more likely to be useful in puerperalfever than one derived from any other source of streptococciexcept puerperal fever itself.

An AddressON

THE TREATMENT OF PULMONARY TUBER-CULOSIS BY HYGIENE.

Delivered before the Cardiff Medical Society onMay 29th, 1902,

BY C. THEODORE WILLIAMS, M.A. OXON.,M.D., F.R.C.P. LOND.,

CONSULTING PHYSICIAN TO THE HOSPITAL FOR CONSUMPTION ANDDISEASES OF THE CHEST, BROMPTON.

MR. PRESIDENT AND GENTLEMEN,-The subject whichI have chosen for my address is perhaps the one mostlargely discussed by the medical profession and laity atthe present moment and it may be said to interest allclasses of the community from the highest to the lowest.My object in choosing this topic is to offer to you some

thoughts and reflections arrived at after a considerable

experience in the treatment of tuberculosis by hygiene andespecially to indicate the best methods of applying this

treatment. What I mean by treatment by hygiene is whatis usually termed open-air treatment, but I prefer the formertitle as more comprehensive and, indeed, more correct, andI thus distinguish it from medicinal and climatic treatment.

Hygienic treatment was probably first systematicallyL practised by Mr. George Bodington of Sutton Coldfield,

Warwickshire, whose reame for consumptives was admirable’

and apparently very successful, though either from beinginsufficiently known or unpalatable it does not seem

to have been largely adopted at the time. LaterBrehmer founded the first sanatorium for consumptivesat Gorbersdorf in Silesia, and introduced his systemof graduated exercises in the way of walking in com-

bination with regulated foods and an open-air life.L Brehmer’s reason for introducing graduated ascents was

that he believed phthisis to be caused by a small heartwhich he thought might become more developed by suchexercises. His views on the etiology of phthisis generallywere remarkable. He held that at the bottom of the mischiefwas the want of food or a small appetite, and that this led tomalnutrition, especially of the heart, which was accom-

panied by an enlargement cf the lungs, and that this want ofproportion between the heart and the lungs was the truepredisposing cause of phthisis. Brehmer lived long enoughto recognise the actual cause of tuberculosis in the tuberclebacillus, but to the end of his life he insisted on theprominent part played by the small heart and the import-ance of increasing its dimensions by exercise. This was amistake on his part, as the small heart is due to that wastingof the organs and tissues which affects the body generally inphthisis and is the result, and not the cause, of the disease.Another point which Brehmer insisted on was that the placeschosen for the treatment of consumption should be immunefrom the disease and he dwelt on the importance ofaltitude in this respect. But since his time we have learnta good many lessons about immunity, and one is that sparse-ness of population has much to say to it and that the preva-lence of tubercle depends largely on the density of population.But in spite of all this Brehmer’s practice was excellent andhe was able to show very satisfactory results. Dettweiler,his disciple, carried on his treatment but diminished theamount of exercise and introduced the rest cure, in which

patients pass the greater part of the day on couches in theopen air, properly protected from rain and wind in liege-halles or shelters. This is the system most in vogue in

Germany now and it is especially adapted to febrile andadvanced cases. Walther at Nordrach modified this systemagain and claims that he follows more closely in the steps ofBrehmer, the chief features of his regime being (1) graduatedexercise in all sorts of weather and indifference to wet andcold ; (2) rest in bed or on the sofa at stated hours andalways in the solitude of the bedroom and not in the societyof the liegehalle; and (3) forced feeding with highly nutritiousfood three times a day. Both these methods combine open-air treatment with a certain amount of stuffing and the greatdifference between them lies in the quantity of exercise

prescribed and of the social intercourse permitted. In both

systems careful observation and- regulation of every detail ofthe patient’s life are inculcated and it may at once be statedthat both systems show a large measure of success.One of the great advantages of the open-air treatment is

that it is suited not only to cases of early and limited

pulmonary tuberculosis which can be arrested by variousmeasures, but that its benefits are also well shown in casesof extensive tuberculisation with cavities, irritable or

quiescent, and above all in pyrexial cases, where owing tothe pulmonary changes the pyrexia is often very persistent.In such patients, after all the antipyretics, pharmacopceialand extra-pharmacopceial, have been exhausted, it is foundthat exposure on a bed or on a light couch by day and bynight to the pure air in a shelter or verandah will in timeprove effective and the pyrexia will gradually subside. Howfresh air of a low temperature affects the thermotaxic andthermogenic nerve centres and thus reduces fever it is noteasy to say-whether by soothing the nerve centres or bystimulating and finally exhausting the combustive processI cannot tell, but the fact I have noted in many instances.I have known patients with active tuberculous disease

accompanied by afternoon pyrexia which had lasted formonths placed on couches in shelters and I have observedthat in most of these cases the pyrexia has subsided aftersome weeks in the open air. Many of these patients werenot in sanatoriums but in their own gardens under the super-intendence of a medical man or a nurse. The open-air life,whether in an institution or in a private house, at first, andespecially in pyrexial cases, causes a feeling of chilliness,but this soon passes off with proper clothing and particularlyif the invalid remains in the prone position with the head

AA 2