The tables

The entry uses the numbers that belong to each passage and says where they come from. This piece sets every count side by side, so that a reader who wants to check can see what each figure is, who gave it, and why the next one differs. All of them were read from the page images of the 1861 book and of the 1905 collected works, because the optical character recognition of the scans garbles tables.

The whole house, 1784 to 1848

Semmelweis’s Table XVII is the “return of strength” of the Imperial and Royal Lying-in Institution from its opening on 16 August 1784 to the end of 1848, one line a year: admitted, died, and the percentage. It is for the whole institution, before and after its division into two clinics. The years that matter most in the entry are these.

1784: 284 admitted, 6 died, 2.11 percent (from 16 August).
1822: 3,066 admitted, 26 died, 0.84 percent. Boër’s last year.
1823: 2,872 admitted, 214 died, 7.45 percent. Klein’s first year.
1827: 2,367 admitted, 51 died, 2.15 percent.
1841: 5,454 admitted, 330 died, 6.05 percent.
1842: 6,024 admitted, 730 died, 12.11 percent. The worst year.
1846: 7,027 admitted, 567 died, 8.06 percent.
1847: 7,039 admitted, 210 died, 2.98 percent. The washings from mid-May.
1848: 7,095 admitted, 91 died, 1.28 percent.

Semmelweis’s summary of the series: from 1784 to 1823 “there occur twenty-five years in which not even one percent” of the women died; Boër, who taught from 1789 to 1822, “in twenty-one years did not lose one lying-in woman in a hundred; for six years the mortality was one in a hundred, four years two in a hundred, one year three and one year four.” From 1822 to 1858 inclusive it was “one year 0 percent, three years 1 percent, six years 2 percent, four years 3 percent, six years 4 percent, four years 5 percent, three years 6 percent, four years 7 percent, five years 8 percent, one year 12 percent.”

Two cautions. The yearly totals of Table XVII do not match the sums of the two clinics in Table I (for 1842, 6,024 against 3,287 plus 2,659, which is 5,946; for 1846, 7,027 against 4,010 plus 3,754, which is 7,764), and Semmelweis does not reconcile them; the one counts admissions to the whole house, which also had a paying department, and the other deliveries on the two free clinics, and they were drawn from different returns. And the deaths are understated in the bad years, as the next section explains.

The two clinics, 1839 to 1846

Semmelweis’s Table I gives the six years from 1841, when the First Clinic had been “set aside exclusively for the instruction of obstetricians.” Skoda’s lecture of 18 October 1849 gives the same series from 1839, “drawn up from official returns,” with a note that the complete separation of students and midwives took place on 19 April 1839. Semmelweis dates the separation to the imperial resolution of 10 October 1840. The entry gives the resolution.

Doctors’ division, deliveries, deaths, percent:

1839: 2,781, 151, 5.4 (Skoda only)
1840: 2,889, 267, 9.5 (Skoda only)
1841: 3,036, 237, 7.7
1842: 3,287, 518, 15.8
1843: 3,060, 274, 8.9
1844: 3,157, 260, 8.2
1845: 3,492, 241, 6.8
1846: 4,010, 459, 11.4
1841 to 1846: 20,042, 1,989, 9.92

Midwives’ division:

1839: 2,010, 91, 4.5 (Skoda only)
1840: 2,073, 55, 2.6 (Skoda only)
1841: 2,442, 86, 3.5
1842: 2,659, 202, 7.5
1843: 2,739, 164, 5.9
1844: 2,956, 68, 2.3
1845: 3,241, 66, 2.03
1846: 3,754, 105, 2.7
1841 to 1846: 17,791, 691, 3.38

Semmelweis says the difference was “in reality far greater,” because “sometimes, when the mortality got out of hand, all the sick lying-in women were transferred from the First division to the General Hospital, died there, and were then entered as dead in the returns of the hospital and not in those of the lying-in house.” Skoda’s footnote says the same: the table “gives the number who died on the division for students as smaller than it really was.” The Second Clinic “never carried out transfers on such a scale.” How many were transferred is not recorded. In the discussion of 18 June 1850 Semmelweis said that since June 1847 no woman had been transferred from the First Clinic except for syphilis, smallpox and the like, “so that the mortality ratio now given is a true and correct one.”

The ratio of the two clinics in 1846 is “five times” in Semmelweis’s text (11.4 against 2.7 is 4.2) and “four times” in Skoda’s. The entry quotes Semmelweis’s sentence and gives both figures.

In the minutes of 15 July 1850 Semmelweis gave the Second Clinic’s mortality as 7.5 percent in 1842 and 5.9 in 1843, “whereas in the year before it was only 2.3 percent and in the year after 3.5.” His own Table I has 3.5 for 1841 and 2.3 for 1844: the minutes have the two years the wrong way round. The entry follows the table.

The months of 1847

From Semmelweis’s text (pages 55 to 56) and Table XVI’s neighbour on page 56. The First Clinic only.

April 1847: 312 delivered, 57 died, 18.27 percent. He had resumed daily dissection and the duty of examining every woman on 20 March.
May 1847: 294 delivered, 36 died, 12.24 percent. The washings began in the second half of the month, on a day he could not remember.
June: 268, 6, 2.38
July: 250, 3, 1.20
August: 264, 5, 1.89
September: 262, 12, 5.23
October: 278, 11, 3.95
November: 246, 11, 4.47
December: 273, 8, 2.93
June to December: 1,841, 56, 3.04

Schürer von Waldheim adds, from the clinic’s reports, that under Breit the mortality had fallen to 1.92 percent in February 1847 and was 3.60 in March, of which Semmelweis served the last ten days.

The September rise Semmelweis blamed on students who had neglected the washing; Hebra’s notice of December 1847 says the same. The October and November figures are the cancer case and the carious knee. The entry tells both.

Hebra’s two counts

Hebra’s first notice, in the December 1847 number of the Zeitschrift der k. k. Gesellschaft der Aerzte zu Wien, gives the result as it stood on 26 November 1847: in April and May, before the measure, “more than 18 deaths to 100 births,” and from then “up to and including 26 November the number of deaths stood to the births as 47 to 1,547, that is, of 100 there died 2.45.” That is not quite the monthly table’s series: June to October in the table is 1,322 deliveries and 37 deaths, and 1,547 is roughly June to the twenty-sixth of November. The two counts agree in substance and were made from the same reports at different dates.

His second notice, April 1848, gives December 1847 as 273 births and 8 deaths, January 1848 as 283 and 10, February as 291 and 2, “and in March no lying-in woman died,” and for “the ten months in which washing with chlorinated lime has been carried out before every examination, of 2,670 delivered only 67 have died, a number that was formerly often exceeded in a month.” The table’s June 1847 to March 1848 is 2,691 deliveries and 68 deaths; Hebra’s ten months probably began in the second half of May.

The year 1848

Table XVI, the First Clinic by month. Births, deaths, percent.

January: 283, 10, 3.53
February: 291, 2, 0.68
March: 276, 0, 0.00
April: 305, 2, 0.65
May: 313, 3, 0.96
June: 264, 3, 1.13
July: 269, 1, 0.37
August: 261, 0, 0.00
September: 312, 3, 0.96
October: 299, 7, 2.34
November: 310, 9, 2.90
December: 373, 5, 1.34
The year: 3,556, 45, 1.27

The Second Clinic in 1848: 3,219 delivered, 43 died, 1.33 percent.

Here is the one real inconsistency in Semmelweis’s own figures. The First Clinic’s 1848 deliveries are 3,556 in Table XVI, in the text on page 56 and in Table XXIII; but on page 108 he writes “in the year 1848, when the chlorine washings were practised the whole year through, of 3,780 lying-in women 45 died, that is 1.19 percent,” and Skoda’s lecture has the same 3,780 and 1.19. The difference, 224, is about the number of women admitted and not delivered on the clinic, or delivered elsewhere and nursed there; he does not say. The entry uses 3,556 and 1.27, the figures of the tables. The 45 deaths are the same in every source.

Haller’s report on the General Hospital for 1848 gives the fall in the First Clinic’s deaths “already in the year 1847, with almost the same number of births, by 283, from 11.4 percent to 5.04 percent,” and says that in 1848 “the mortality ratio became equal to that of the Second Clinic, indeed by chance 0.1 percent more favourable.” Haller’s “283” is 459 less 176.

The first months of 1849

From the text on page 57: January 1849, 403 births and 9 deaths, 2.23 percent; February, 389 and 12, 3.08 percent; March, 406 and 20, 4.9 percent. Semmelweis left on 20 March 1849 and Carl Braun took over as assistant that day. Skoda, in October 1849, gave the two clinics’ deaths “up to the beginning of September” 1849 as 60 on the students’ division and 76 on the midwives’.

The newborn

Semmelweis’s counts of infants who died on the two clinics, which he gives as evidence that the mothers’ blood carried the disease to the child before birth:

1846, no washings: First Clinic 235 of 3,533 (6.65 percent); Second 86 of 3,398 (2.53).
1847, washings for the last seven months: First 167 of 3,322 (5.02); Second 90 of 3,139 (2.86).
1848, washings all year: First 147 of 3,496 (4.20); Second 100 of 3,089 (3.23).

The infant deaths of 1848, he adds, “were not caused by puerperal fever.” The infant totals (3,533 in 1846, 3,496 in 1848) are not the same as the maternal totals (4,010 and 3,556), because stillbirths and twins are counted differently; he does not explain.

Routh’s approximations

Routh, writing in London in May 1848 from what he had seen on the wards, gave the First Clinic’s deliveries as “from 250 to 300, mean 275” a month, with “about 30” deaths ordinarily and “about 70” occasionally, “1 in 9.16” and “1 in 3.92”; and the Second’s as “about 7 to 9, mean 8” and occasionally “about 3,” “1 in 34.37” and “1 in 91.66.” He added: “In one particular year, indeed, out of some 3000 cases delivered, about 500 died in the first division” (the scan’s OCR reads the figure as 8000; the German translation of 1905 has 3000, which fits the year). “While in the second division, out of the same number delivered, the deaths frequently did not exceed forty. These numbers are only approximative, as for obvious reasons the statistics are kept secret.” The year he had in mind was 1842, when the First Clinic lost 518 of 3,287. His “seven per month” after the washings is the Second Clinic’s ordinary figure, and matches the First Clinic’s June to August of 1847 (6, 3, 5). The backlog’s “about one in ten and one in thirty” is Routh’s ratio.

The two clinics, 1847 to 1858

Table XXIII, with Semmelweis’s footnotes on the assistants: Semmelweis from 20 March 1847 to 20 March 1849; Carl Braun from 20 March 1849 to the summer of 1853; Gustav Braun, Carl’s brother, assistant from 1853, and acting professor from April to December 1857; Carl Braun professor from 1858. Births, deaths, percent.

First Clinic (doctors):

1847: 3,490, 176, 5.04
1848: 3,556, 45, 1.27
1849: 3,858, 103, 2.66
1850: 3,745, 74, 1.98
1851: 4,194, 75, 1.79
1852: 4,471, 181, 4.04
1853: 4,221, 94, 2.22
1854: 4,393, 400, 9.10
1855: 3,659, 198, 5.41
1856: 3,925, 156, 3.97
1857: 4,220, 124, 2.93
1858: 4,203, 86, 2.04
Twelve years: 47,935, 1,712, 3.57

Second Clinic (midwives):

1847: 3,306, 32, 0.96
1848: 3,219, 43, 1.33
1849: 3,371, 87, 2.58
1850: 3,261, 54, 1.65
1851: 3,395, 121, 3.56
1852: 3,360, 192, 5.71
1853: 3,480, 67, 1.92
1854: 3,396, 210, 6.18
1855: 2,938, 174, 5.92
1856: 3,070, 125, 4.07
1857: 3,795, 83, 2.18
1858: 4,179, 60, 1.43
Twelve years: 40,770, 1,248, 3.06

His comment: the mortality “has indeed fallen since the introduction of the chlorine washings, in mid-May 1847, by 6.35 percent on the First division and by 0.32 percent on the Second. But the mortality is 2.30 percent greater on the First division and 1.70 percent greater on the Second than in the year 1848, when the chlorine washings were supervised by me; although I too did not achieve the smallest possible mortality, for reasons I have described in the proper place.”

The 1854 figure of 400 is the one he threw at Gustav Braun in the open letter to Siebold: “I am of the opinion that puerperal fever arises in consequence of an infection, and in the year 1848 I sent 45 lying-in women to the mortuary. Gustav Braun is of the opinion that puerperal fever is of epidemic origin, and he, with his ignorant pupils, in the year 1854,” sent 400.

His own arithmetic of the preventable

In the open letter to Spaeth he set a ceiling of one death in a hundred for “unpreventable self-infection,” which he thought too high, and counted everything above it as “preventable infection from outside.” On the two Vienna clinics in the ten years after 1847 that gave him 1,924 preventable deaths: 1,083 on the First Clinic and 841 on the Second, where 1,183 had died of 34,245 delivered. “And in this massacre you, Professor, are a participant.” The 34,245 and 1,183 are the Second Clinic’s 1848 to 1857 from Table XXIII, less a few; the arithmetic is his.

The 1850 lecture’s summary figures

The minutes of 15 May 1850 give the three-year result as a mortality “which was otherwise 8.3 percent of the lying-in women on the First Obstetrical Clinic” become “2.3 percent,” and for the two clinics together: before 1 May 1847, “the period of the flowering of pathological anatomy,” 5.7 percent; from then “to the end of April of this year,” 2.2 percent. The minutes do not say how the 8.3 was computed; it is lower than the First Clinic’s 1841 to 1846 average of 9.92 in Table I.

Pest, 1851 to 1858

The lying-in ward of St Rochus’s Hospital, where Semmelweis was honorary head from 1851, took in women in labour only in the vacation months of the university’s school year, the rest going to the university clinic. His count: 121 births in the vacation of 1850 to 1851, 189 in 1851 to 1852, 199 in 1854 to 1855, 126 in 1855 to 1856 (the two years between are unreadable in the scan), 933 in all; 24 deaths, of which 8 from childbed fever, “0.85 percent,” the other 16 from diseases the women had been treated for before labour. One of the eight was examined “by a surgical registrar after he had just made the post-mortem of a man who died of a gangrenous lower leg.”

At the university clinic in Pest, from 1855, the mortality in the school year 1856 to 1857 and in 1857 to 1858 is given to two decimals which the scan does not render; the text says 31 women died in the first of those years and 24 in the second, and blames “the sheets and the atmospheric air” for both.

The Second Clinic, 1861 to 1863

From Spaeth’s lecture of 16 January 1863 as Schürer von Waldheim summarises it: from 1 October 1861 to 1 January 1863, 4,153 women delivered on the Second Clinic and 212 dead of puerperal processes, 5.1 percent; within that, from 1 October 1861 to 2 February 1862, of 1,127 newly delivered, 236 fell ill and 148 died, 13.2 percent; in November 1861 the morbidity reached 30.3 percent and the mortality 11.25. Of the street births in that epidemic “only one fell ill of childbed fever and none died.”

What the disagreements amount to

None of them touches the shape of the result. The First Clinic lost between one in six and one in nine of its mothers in the months before the basin and about one in eighty in the first full year after it; the Second Clinic, admitting on the alternate days, lost about one in thirty-five before and about one in seventy-five after; and when the supervision changed the First Clinic went back, within five years, to one in eleven. Whether 1848’s denominator is 3,556 or 3,780 moves the First Clinic’s rate from 1.27 to 1.19. The transfers mean the bad years were worse than printed, which makes the fall larger, not smaller. The one number in this whole story that nobody disputed at the time was 45.

Journal of Experiments