Smoking and Carcinoma of the Lung: Preliminary Report
1950 · paper · Biomedicine
A preliminary report on the link between smoking and lung cancer, prompted by the steep rise in lung cancer deaths in England and Wales. The supplied text covers background and possible causes, and does not reach the findings.
Checked: PDF from europepmc.org, fetched 4 Oct 2026. Page numbers are those of that PDF.
1 finding
Reference or label The Table VII Probability Test note '(Women smoking 15 or more cigarettes a day grouped together)' refers to daily consumption, but Table VII is classified by total cigarettes ever smoked. The female χ2=12.97 matches pooling the 150,000+ columns.
- p. 5 Table VII, Probability Test column, female rows
(Women smoking 15 or more cig- arettes a day grouped together)
Judge's note: The page 5 image shows the same wording as in Tables V and VI, but Table VII's columns are 365, 50,000, 150,000, 250,000 and 500,000+ cigarettes ever smoked. Pooling the women's 150,000+ columns (lung 5+7+0=12; controls 3+1+0=4) gives a 2x3 table with chi-square 12.97 on 2 df, which matches the printed χ2=12.97, n=2. The note was evidently copied from the earlier tables.
Second check: another model agreed
Report a problem with this finding
7 minor (typos and the like; most checked by the judge only)
Substance In the Fig. 1 MEN panel, the control bar for 1-4 cigarettes is labelled '9', but 55/649 = 8.5%, which rounds to 8. The paper says the figure percentages are based on all patients in each group.
- p. 4 Fig. 1, MEN panel, control bar at 1-4
9
Judge's note: Page 5 text: 'In the figure the percentages are based on the total number of patients in each disease group, smokers and non-smokers alike.' Every other Fig. 1 label on page 4 matches the Table V count divided by 649 (men) or 60 (women): 4, 0.3, 5, 45, 39, 29, 30, 11, 21, 2, 5, 53, 32, 20, 12, 17, 32, 10, 15, 0, 10. Only the 1-4 control label for men disagrees, by about half a point.
Second check: another model agreed
Report a problem with this finding
Substance Table VII prints male controls in the 250,000 Cigs.- column as '179 (28·9%)', but 179/622 = 28.8%. The row's percentages then sum to 100.1.
- p. 5 Table VII, male control patients, 250,000 Cigs.- column
(28·9%)
Judge's note: The page 5 image shows 28·9%. The other cells in the row are correct with denominator 622 (5.8, 30.5, 29.3, 5.6). This is a rounding slip of 0.1 point.
Second check: another model agreed
Report a problem with this finding
Substance Table X: the italic expected numbers for 'Gastro-intestinal disease (328)' (55.7, 32.3, 130.2, 75.8, 34.5) sum to 328.5. Those for 'Other diseases (215)' (36.6, 21.1, 86.0, 48.9, 22.1) sum to 214.7.
- p. 6 Table X, Gastro-intestinal disease (328), italic expected row
Gastro-intestinal disease (328)
Only the authors can say what is right here.
Judge's note: From the page 6 image, the expected rows for Cancer (718.0), Respiratory (335.0) and Cardiovascular (166.0) sum exactly to their group totals, as the footnote's standardisation method requires. Rounding five 1-decimal values can shift a sum by at most 0.25, so the GI (+0.5) and Other (-0.3) rows are inconsistent beyond rounding. The column sums of the expected values still match the observed totals within rounding, so a value has probably been misplaced between these two rows. Only the authors can say which cell.
Second check: another model agreed
Report a problem with this finding
Wording The place-of-residence statistic '(χ2 =31.49; n=5; P<0.001)' and the Table II residence counts (330/377, 203/231, 23/16, 95/54, 43/27, 15/4) do not correspond under the declared test: Pearson chi-square from these counts is 27.49.
- p. 3 Text below Table II
(χ2 =31.49; n=5; P<0.001)
- p. 3 Table II, Place of residence
330
Only the authors can say what is right here.
Judge's note: From the page 3 image, the residence counts are 330, 203, 23, 95, 43, 15 for lung-carcinoma patients and 377, 231, 16, 54, 27, 4 for controls; both columns sum to 709. With equal group sizes, the expected count in each cell is half the row total. The Pearson chi-square with 5 df comes to 27.49, not the printed 31.49. The same method gives 1.617 for the printed social-class value of 1.61. P<0.001 holds under either value. Only the authors can say whether the statistic or a count is misprinted.
Second check: another model agreed
Report a problem with this finding
Wording The same residence difference is called 'large' ('The diffeience in place of residence is, however, large') and, a paragraph later, the control group 'differs slightly with regard to the parts of England from which the patients were drawn'. The Discussion says the groups 'were not comparable with regard to their places of residence'.
- p. 3 Results text after Table II, left column
The diffeience in place of residence is, however, large
- p. 3 Results, right column, top
but differs slightly with regard to the parts of England from which the patients were drawn
- p. 8 Discussion
the carcinoma and the control patients were not comparable with regard to their places of residence
Judge's note: Page 3 shows both phrases one paragraph apart; page 8 shows the Discussion wording, and page 6 says the groups were 'not wholly comparable'. 'Slightly' may describe the practical consequence rather than the statistical size of the difference, but as worded the descriptions of one comparison are inconsistent. This is minor wording only.
Second check: another model agreed
Report a problem with this finding
Wording 'Thus Table XIV shows clearly, and for each age group, ... that the risk ... increases steadily as the amount smoked increases', but the 35- row has 43 then 41 and the 55- row has 844 then 600.
- p. 8 Discussion, paragraph after Table XIV introduction
Thus Table XIV shows clearly, and for each age group,
- p. 8 Table XIV, rows 35- and 55-
844
- p. 8 Discussion, paragraph introducing relative risks
Thus Table XIV shows clearly, and for each age group, the conclusion previously reached-that the risk of develop- ing carcinoma of the lung increases steadily as the amount smoked increases.
- p. 8 Table XIV, rows 35- and 55-
43 | 41
Judge's note: In the Table XIV crop, the 25- row and a few other cells (e.g. 77, 12, 2,000) are in italics (<5 cases), but 43/41 in the 35- row and 844/600 in the 55- row appear to be roman. So the claim of a steady increase 'for each age group' overstates the table slightly. The overall trend is clear; this is a wording point.
Second check: another model agreed
Report a problem with this finding
Wording 'are 3, 10, 20, 35, and, say, 60-that is, the mid-points of each smoking group': for the 25-49 group the mid-point is about 37, not 35.
- p. 8 Discussion, relative-risk paragraph (right column)
are 3, 10, 20, 35, and, say, 60-that is, the mid-points of each smoking group.
- p. 8 Table XIV column header
25-49
Judge's note: The page 8 image shows the column headers 1-4, 5-14, 15-24, 25-49, 50+. The values 3, 10 and 20 fit class mid-points, but (25+50)/2=37.5 and (25+49)/2=37. Since the figures are explicitly speculative and approximate, this is a minor descriptive inaccuracy.
Second check: another model agreed