Tobacco Documents Online

Vl, BRITISH MEDICAL JOURNAL tr,t LONDON SATURDAY FEBRUARY 7 1959 LABORATORY CONTRIBUTIONS TO THE TOBACCO- - CANCER PROBLEM* BY .; ERNST L WYNDER, M.D. Head, Section of Epidemiology, Division of Preventive Medicine, Sloan-Kettering Institute, New York ; Associate Professor of Preventive Medicine, Sloan-Kettering Division, Cornell University Medical College, New York The sum total of evidence linking smoking to cancer of the respiratory tract is based upon different types of evidence: presumptive, epidemiological, pathological, animal, and chemical. All of the evidence so far established demonstrates smoking to be a carcinogenic factor. It is now our task to brin.- the problem posed by this association to a successful solution. The present report represents an evaluation of the con:ributions that laboratory research is making in this field. The importance of laboratory work is not to prove that smoking is a cause of cancer in man. Such proof can only come from human epidemiological investiga- tion. Laboratory research can, however, contribute to, and give a logical explanation for, the human 41 findings. Just as an animal experiment cannot disprove . a f agents responsible for the activity in animals are also responsible for the human activity. In view of the many similarities established for tumour growth in animals and man, such an assumption, though it cannot be proved, stands on a firm foundation. Most cancer researchers would surely agree that once a carcinogen has been identified in a given material suspected to be active for man and proved to be active for animal tissue, particularly when this is demonstrated for several species of animals, such an agent should, if at all possible, be reduced or eliminated from man's environment. It is along these lines that laboratory research, as it applies to the tobacco-cancer problem, has its greatest significance. It is now our purpose to review the methods followed in respect to this work and the results already achieved. that a given factor causes cancer in man because of possible species differences, so, by itself, an experiment cannot prove a given agent to be carcinogenic to man. It is primarily as a corollary to the human findings that the animal experiment has its significance. The basic tasks of laboratory research, which are of a biological and chemical nature, are to identify the specific agents in a given product that produces cancer and to devise ways and means whereby such agents can be reduced or removed. In so doing, we can only assume that the Methods of Study J• 0 Since the primary purpose of the biological study is to establish the activity of the agents suspected to be carcinogenic, the test site is perhaps less important than *Presented before the Seventh International Cancer Congress in London, July 11, 1958. is generally considered. In choosing the test site, one must be sure not only to use a site which is not too sensitive to tumour formation, but also to avoid one in which tumours cannot be produced even with very potent carcinogens. In general, it would be advantageous to use the type of tissue similar to the one involved in man. In view of these considerations, the subcutaneous tissue of mice would be a less useful site because it does not yield epithelial tumours and also because it has been shown to be quite sensitive to a large variety of substances. On the other hand, the lungs of mice would not represent a good test organ, since, even upon inhaling high doses of potent polynuclear hydrocarbons,e'S it has been difficult to produce lesions in the bronchus in mice. The skin, on the other hand, is a satisfactory site not only because of ease of application, but also because it represents a type of tissue similar to the epithelial tissue of the respiratory tract. An important factor when testing a product to which man is exposed is to test this product under a condition similar to that under which man is exposed. Thus we should smoke tobacco in a manner simulating human smoking habits, and should not distil the tobacco smoke in a closed container. Another important principle is that, when testing a substance suspected to be only weakly carcinogenic, the substance should be applied in maximum concentration over a maximum period of time. With these considerations on methodology we shall review the actual experiments already completed. BIOLOGICAL DATA Lung Studies A number of experiments have been conducted exposing mice to cigarette smoke. As could be expected from similar experiments with pure carcinogens, it is most difficult to produce bronchiogenic lesions in this manner. The method is particularly difficult with tobacco smoke because if the concentration of the smoke is too high animal mortality is too great. Campbell, and later Essenberg, have succeeded in producing pulmonary adenomas in susceptible mice by exposing the animals to varying concentrations of cigarette smoke.' 16 Lorenz obtained negative results.S2 In a more recent and detailed study, the Leuchtenbergers found that there is an increase in hyperplasia, metaplasia, dysplasia,-and carcinoma in situ when mice are exposed to cigarette smoke for a relatively short period of time.31 These experiments are of interest because they show 5118

318 FEB. 7, 1959 CONTRIBUTIONS TO TOBACCO-CANCER PROBLEM Before the chemist could proceed to identify the active' materials, biological study had to be conducted to determine the particular components of the total tar in which the majority of the activity is located. In a large-scale study, summarized in Fig. 1, we have shown MEDICAL rOURNAL the same types of early bronchial changes as demon- strated by Auerbach and others' 1= 22 in human lungs.2s Using a more direct approach, Rockey applied condensed cigarette smoke to the trachea of dogs for about 11 months, noting severe metaplasia of the epithelium.a0 Blacklock, injecting tobacco smoke condensate together with tubercle bacilli into the hilum of rats, was able to produce two carcinomas in the eight rats treated in this way.4 Though inhalation studies, therefore, have so far not produced any actual bronchiogenic carcinomas in the experimental animal, a result which could not be expected because of the toxicity of high doses of tobacco smoke and in view of the fact that this has been difficult even with high concentrations of polycyclics, the available evidence has nevertheless indicated an abnormal reaction of the pulmonary and bronchial tissue to tobacco smoke and in one instance the production of carcinoma in this tissue when the smoke condensate was directly applied. Skin Studies There are surprisingly few experiments dealing with the production of skin cancer in animals upon applica- tion of tobacco smoke condensate in view of the attention given to the problem. In fact, until 1953 no study had been done with condensed tobacco smoke. At that time we published our first report showing the production of 44% cancers and 59% papillomas among CAF, mice which had been painted with a 50% tar-acetone solution three times a week 45 In 1955 we reported'~ the production of cancers in two additional strains of mice, Swiss and C;,. Since this time we have reported positive results on yet another mouse strain.1e In the meantime, Hammer and Woodhouse, as well as Passey, reported their inability to produce skin cancer in mice with tobacco tar.21 " However, as will be shown subse- quently, the negative results are not necessarily due to differences in British and American tobaccos but rather to the fact that the tar was applied at a subthreshold level. The fact that tobacco tar is carcinogenic to mouse skin has since been confirmed by Sugiura, by Bock, by Orris, and by Engelbreth-Holm.3 15 38 44 -1-heze studies leave no doubt that tobacco smoke condensate is CIGARETTE TAR FRACT/ONAT/ON- RELAT/VE BIOLOG/CAL ACT/V/TY WHOLE TAR 1009. ++++ MlCI2INSOL. 14g. MBCI2 SOL. 86g. + I H2504 ACIO + NEUTRAL 60g. ++++ Na2co3 Bunsa BASIC TAR 149. NICOTINE FREE BASIC 6.29. + NEUTRAL TAR ACIDIC TAR 439. 12g +++ + 1 I l HEXANE CARBON TET. BENZENE ETHYL METHANOL PYRIDlnE ACETATE 6.0g. - 1.7g. 6.0g. 21.Sg. 6.2g. 1.0g. + ++++ FIG. 1.-The plus sign indicates relative values of carcinogenic activity. c ma a r that the majority of the active fractions of tobacco smoke condensate are in the fraction which is eluted with carbon tetrachloride from the neutral tar 49 This fraction, representing only 1.7% of the totall tar, produces 100% cancer in animals when applied in 10% concentration. We did observe some activity in other fractions. However, it cannot be said whether this is a result of independent carcinogenic substances or whether it is the consequence of unsatisfactory chemical separation. Present data suggest that in the basic portion of the tar, where polynuclear substances are not thought to remain, there are at least cocarcinogenic elements.49 This is believed to be the case not only because of the high hyperplastic reaction obtained with this material, but also because it increased the tumour yield when added to the neutral tar. However, we ennclnrle at thi ti th t th . d r 0 4.

a u ?~ 4 < t d ~-r I FEB. 7, 1959 CONTRIBUTIONS TO TOBACCO-CANCER PROBLEM Ba<Ttsp 319 .YDtC:AL )oURNAL thrysene,S3 311 3: 4: 9: 10-dibenzpyrene,a' and 3:4: 8: 9- dibenzpyrene.33 14 The last-named has also been tested by Buu-Hoi, but has relatively little biological activity for the skin.25 Recently, Hoffmann, and also Van Duuren, have identified 3:4-benzfluoranthene, which we have proved to be carcinogenic to mouse skin." as Additional higher aromatic polycyclics identified but not yet tested for carcinogenic activity include benz(mno)- fluoranthene, 10:11-benzfluoranthene, and 11:12-benz- fluoranthene.=3 as Chemical work done at present in our laboratory is directed toward determining additional polycyclics in the various tobacco fractions found to be carcinogenic. In Table I, as most recently completed by Dr. Hoffmann, TABLE I Polycyclic Hydrocarbons 3:4-Benzpyrene t:2:5:C.Dibenzanthracene . . 3:4-Benzfluoranthene 10:11- 1:2-Benzpyrene 1:12-Benzperylene .. .. .. 1:2-Benzanthracene . . .. .. Chrysene Alkylchrysene Fluoranthene.. .. .. .. Alkylfluoranthene .. .. .. Pyrene Alkylpyrene .. Perylene It:12-Benzfluoranthene Benzo(mno)fluoranthene .. 2:3-Benzfluorene .. Anthracene . .. .. .. Phenanthrene ~ p.p.m. Fraction B 0.57 . 10° 0•24 . 10' 0055 . 10= 0-18 . 102 0055 . 10s 0.09 . 10° 1-1 +. 102 0-55 . 102 0.93 , 102 + 4 0 . 10s 1•6 . 101 0•03 . 10= 0•1 . 10' + + + p.p.m. Fraction C 1 •14 . 10' 0-01 . 106 0•73 . 10s 0 08 . 10s + 0 99 . 10' + + + 0.1 . 105 Fraction B is the carbon tetrachloride eluate of the neutral tar (see Fig. 1). Fraction C is biologically the most active fraction of the 880° C. pyrolysate of a hot hexane extract of cigarette tobacco. we show the identification of higher aromatics present in the carbon tetrachloride fraction of the neutral tar. Even though at present we may still not have identified all of the polycyclics responsible for the total activity of this fraction, the activity of this fraction is largely due to polycyclics. There obviously remain other polycyclics still to be identified, a project which may be of greater academic than practical importance, since it may be assumed that polycyclics are produced in the same manner. These tables also show the identification of polycyclics in Fraction C, representing one of the ten subfractions of the 880° C. pyrolysate of hexane extracted tobacco which, in 0.01 %'concentration, proved to be biologically active.s' It is of interest that, even though a whole range of polycyclics was identified in the 880° C. pyrolysate (Table II), only Fraction C in a 0.01 % concentration proved to be biologically quite active, while Fraction B had very minor activity. This TABLE IL-Polycyclic Compositiort of 880° C. Pyrolysate of Hot Hexane Extract of Cigarette Tobacco Determined Spectro- photometrically Pyrolysate Fraction A Mixtureofaliphatic hydrocarbons, naphthalene, mono-substituted aromatics (yellow oil), phenanthrene, 4-methylpyrene, anthraceno, pyrene, fluoranthene, and a mixture of unknowns. B Unknown (alkyl anthracene), 1:2-cyclopentanophenanthrena, 1:2-benzanthracene, chrysene, acenaphthylene (trace of 3:4- benzpyrene). C Perylene, 1:2-benzpyrene, 3:4-benzpyrene, 1:12.benzperylene (total in Fractions C and D), and other polycyclics (Tal:1e I). D 1:12-Benzperylene and other polycyclics. E 1:2:3:4-Dibenzpyrene,3:4:8:9-dibenzpyrene,3:4:9:10-dibertzpy- rene, and other polycyclics. F 1:2:7:8-Dibenznaphthacene,2':3'-naphtho-3:4-pyrene,coronene, and other polycyclics. G Coronene, 1: 2:5:6-dibenzanthracene, and other polycyclics. H Unknown, 3:4-benzpyrene derivative and other polycyclics. I A mixture of several unknown polycyclic compounds. would suggest that the majority of the carcinogenic polycyclics present in tobacco tar chromatograph in the region of benzpyrene. The identification of 3:4 benzfluoranthene as an active carcinogen represents a case in point. Additional work in which Dr. Hoffmann is engaged concerns studies with radioactive benzpyrene and 1:2:5:6-dibenz- anthracene, and is designed to determine the effectiveness of our chemical separation schemes in removing. the benzpyrene present in the total tar in the final solutions. Though the higher polynuclear substances are the only carcinogens present in tobacco, they are the major carcinogenic components, and may be regarded as a standard olt which to predict the carcinogenic activity of any type of tobacco smoke condensate. Preventive Approaches Having determined the major tobacco carcinogens, it now remains to be considered how the carcinogenic activity of tobacco smoke could be most effectively reduced. Dose-Response Studies In view of the established principle of carcinogenesis that the higher the dose the greater, up to a given point, the tumour yield, it became pertinent to establish this factor for the ex- perimental animal.'0 In Figs. 2 and 3 we summarize the data in this respect. They indicate that there is a minimum as well as an optimum level at which tobacco smoke condensate produces cancer in t h e experimental animal, and that the minimum level is about one-third of the optimum level. so 70 60 0 sa ll l~ :o ,e J 0 1 GMS 2 3 4 5 8 7 e 9 ro CIGORETrE TAR E[PoSVRE IGYS I/r.pVSE/rR GROURSID< II ffiYID ID IIII ¢IIZ FtG. 2.-Percentage of papillomas by 18 months. 50r- 40 30 20 lo ~1 1 0 GMS 2 3 4 5 6 7 8 9 10 CIGARETTE TAR EXPOSURE (GMS.) /MOUSE/YR GROUPS YII 27 >?BYYI III 1g%11 IIZI % ' Fl:a. 3.-Percentage of cancers by 18 months. These studies are of academic importance in that they demonstrate the reasons why certain investigators may have been unable to produce tumours with tobacco products. More important, they are of practical significance in that they show that if tar exposure is reduced below a certain point the rate of tumour formation, at least in the experimental animal, is

320 FEB. 7, 1959 CONTRIBUTIONS TO TOBACCO-CANCER PROBLEM greatly reduced. In this respect they are similar to the epidemiological studies which show the same relationship. In view of these conclusions it becomes of obvious practical importance to determine ways by which the tar content of the tobacco smoke condensate can be reduced. Tobacco Types Considering the possibility that various tobacco types might differ in the production of carcinogenic substances, we undertook a study of cigarettes made of pure Burley, Maryland, Turkish, and Virginia tobaccos.'= The results of these studies are summarized in Fig. 4, and indicate 0 .o !D 0 L T EURLEY G °pF • Bwiss MARYLAND TURRISM VIRGINIA TYPE OF TOBACCO FIG. 4.-Percentage of cancers at 18 months, with different types of tobacco. no significant varia- tion in carcinogenic activity, even though there are obvious variations in nico- tine content a n d thus in the base- f r e e fractions of these tars. T h e s e comparisons could be made only for the base-free portion of the tars, since the nicotine content of the Burley tobacco is too high for biological testing. We conclude at the present time that tobacco selection, though it can greatly influence the nicotine content and certainly can also influence the total tar content of the condensate, will not significantly influence the carcinogenic activity of the tar on a gramme-to-gramme basis. Filter Cigarettes In view of the points emphasized by the dose-response studies, we were interested in conducting studies with filter cigarettes. As summarized in Fig. 5,'1 these studies . indicate that on a to UR •U.IiRIr,C R,qu~a, UN • Un1~it,r,E F nq fR • F~R,rta Requ~ar FR.F.Rw,O F~•q O C1F SwisS gramme - to-gramme basis the carcino- genic activity of tar obtained from filter cigarettes is similar to that of unfiltered } cigarettes. Therefore it, is established that ° a mechanical filter URI UvII UKa FRI FFIa FRI FN$ FKII TYPE OF CIGARETTE cannot selectively FIG. 5.-Percentage of cancers at 18 remove the carcino- months, with different types of cigarette. genic materials from tobacco smoke, which, knowing the physical make-up of tobacco smoke, could have been predicted. A filter thus serves its purpose not becau'se it removes certain components of tobacco tar selectively, but because it can lower the total tar content of the smoke. Any filter which does not fulfil this requirement is not useful. In the past some tobacco manufacturers, while employing a fairly efficient filter, used high tar-yielding tobaccos. Such smoking products are misleading to the consumer. Tar reduction can be most effectively achieved by a combination of efficient filtration and proper tobacco selection. Data recently reported showing the tar content of filtered and unfiltered cigarettes as currently smoked in the United States indicate that to-day there is a cognizance of this principle by at least some of the tobacco manufacturers, in this country.36 As indicated BRmsa MEDICAL IOURNAL TABLE Itt 1957 I 1958 Change Tar content in mg. ojsome 85-mm. filter rigarettes' Filter A.. .. 32 6 29-1 B . .. I 30•4 17 9 C .. .. .. 30-2 27•7 D .. .. .. 38•5 24•9 E .. 36-3 17-7 F(70 mm.) 25•6 15.1 Tar content in mg. ojsome 73-mm. regular cigarettes Brand A .. .. .. 31-0 28-7 B .. .. .. 31 •5 28•6 C .. .. 32-7 30.6 D .. .. .. 35•3 28•9 E .. 30-9 22•4 „ F (85 mm.) .. 38•6 37•1 -48% -35% -5l% -41% -7% -9% -~g j -41% -4% t • Foster D. Snell.xa in Table III, a marked reduction in some of the major cigarette brands in the United States has taken place. This movement is to be encouraged, and it is hoped that before long all of the tobacco manufacturers in the United States and elsewhere will follow suit. While such a move will not prevent a smoker from developing lung cancer, present evidence indicates that it will reduce his chances of developing this disease. Pyrolysis Studies In biological experiments we have shown that an extract of tobacco is only weakly carcinogenic compared with tobacco smoke condensate.'B There is present in unburned tobacco a very small amount of some higher aromatic polycyclics which are apparently formed during the curing process.10 17 16 However, it is clear that the majority of the higher polynuclear substances are formed during the combustion processes of tobacco. We have set out to undertake a series of studies to determine the temperature ranges at which the majority of the carcinogens are formed. We have pyrolysed hot-hexane-extracted tobacco at temperatures ranging from 880° C. to 560° C. and found that the formation of carcinogens is related to the burning temperature. It does not appear to be so m u c h related to the pre- sence or absence of oxygen, since the activity of the 880° C. pyrolysate in nitrogen and with t h e addition of oxygen was similar. However, we found that when the temperature w a s reduced below 700° C. the biological activity was greatly reduced (Fig. 6).53 eo 60 50 40 30 20 10 0 FIG. 6.-Tumour formation in mice upon application of pyrolysis products of hexane extract of tobacco obtained We were interested at different temperatures. in investigating this problem from a number of different aspects. We smoked cigarettes with a high and a low puff volume in an effort to determine whether this would alter the carcinogenic activity. In view of the fact that maximum temperatures reached in these cigarettes are quite similar, 884° C. ± 30° C., we did not expect a variation in activity, and, indeed, none was found.'- It is of interest in this respect that the cigarettes smoked with a high puff volume yielded more tar. However, all our experiments are based upon a gramme-to-gramme comparison. We 1 0 11 If . ~ >1 I I r

>• M 0 s w 1 FEB. 7, 1959 CONTRIBUTIONS TO TOBACCO-CANCER PROBLEM BArmH 321 MEDICAL IOURNAL also set out to test whether cigarettes smoked halfway or to the butt end would show different activity, since repyrolysis of condensed tar might produce more carcinogenic material. However, the results again showed a similarity in biological activity even though the tar yield of cigarettes smoked to the very butt end is obviously greater than of those smoked halfway down.•53 Finally, we have studied the comparative activity of cigar, pipe, and cigarette tars. We found cigar and pipe tar somewhat more active, which, we believe, is due to the fact that cigar and particularly pipe tobacco burns at a high level for a longer period of time than does cigarette tobacco, even though the maximum temperature of cigarettes is higher than that of pipes. These temperature studies have been reported in detail by Touey and Mumpower, and are summarized in Fig 7.28 °' In burning at a high temperature for a longer rany..'CC iooo foo f00 roo mo _ A Cqanrres ah" hyu U these observations it would be most difficult to remove any given substance from tobacco in the absence of which no polycyclics could be formed. We believe, therefore, that even though the different components in tobacco may vary in their relative susceptibility to form higher aromatic polycyclics, a removal of certain substances from the tobacco itself would not be a practical way of reducing its carcinogenic. activity upon being smoked. Practical Preventive Measures The practical preventive measures as derived from completed laboratory work fall into the following categories. 1. Lowering of Tar Content in View of Studies on Dose-Response Levels.-This can be attained through effective filtration and tobacco selection. The greater the decrease in tar content of a given cigarette the lower the liability to cancer development. This is a practical step which can be undertaken by the tobacco industry without delay. 2. Temperature Reductants. - We are currently engaged in a study of a number of substances, including aluminium products, to determine whether the temperature of the tobacco during smoking can be lowered sufficiently to influence the formation of polynuclear substances.5' A number of suggestions have been made to cool the main stream of the smoke. However, since the carcinogens undoubtedly are formed 1 0 11 in the burning process, it is here that we must concentrate FIG. 7.-Potentiometer graphs (Touey and Mumpower'e eS). period of time the combustion may be more complete in cigars and pipes. Obviously the formation of the carcinogens from organic material is not only a consequence of maximum temperature, but also of the duration of contact with a given temperature level. Present evidence strongly suggests that modification of temperature levels, if achieved, could influence the formation of carcinogens in tobacco. Study of Precursors By studying precursors we planned to determine whether there were any components in tobacco smoke condensate which would be particularly susceptible to the formation of higher aromatic polycyclics. We washed tobacco with hot hexane and smoked the extracted tobacco. Of immediate interest is the fact that, though only 5.4% of the tobacco by weight was removed, the tar yie;d of this cigarette was 35% less than that of an ordinary cigarette.'' However, on a gramme-to-gramme basis, one of the two experiments showed a somewhat decreased activity and the other showed no decrease in activity from regular tar. Therefore, at present we must conclude that this method cannot effectively reduce the carcinogenic activity of tobacco tar. Lindsey had previously shown some reduction in benzpyrene content of hexane-extracted tobacco.`° However, our studies do not show a reduction of benzpyrene in hexane-extracted tars. Lindsey has shown that a large variety of agents present in tobacco can produce higher aromatic poly- cyclics when pyrolysed.' 9 Lam has pyrolysed some of the sterols present in tobacco and has identified higher aromatic polycyclics.=8 We have shown this pyrolysate to be biologically active on mouse skin.'' In view of our efforts. 3. Modi fication of Pyrolysis.-Through the use of a variety of catalysts we are currently engaged in determining whether the polynuclear content of tobacco smoke condensate can be reduced." The idea of catalysts, which is useful in the petroleum industry, may be less applicable in the case of tobacco because of shorter contact t:me. However, work completed so far suggests that the polynuclear content can be altered. It also seems to affect the proportion of different polynuclear substances. These studies are still in the preliminary stage, and it remains to be determined through combined biological and chemical investigations. whether there is a particular catalyst or group of catalysts which could reduce in a practical fashion the carcinogenic activity of the tobacco smoke condensate. Conclusion In summary, we have reviewed the work being conducted in various laboratories throughout the world, .and particularly in our own laboratory, relating to the tobacco-cancer problem. We have stated the purpose of the laboratory experiment, the direction in which it must go, and have emphasized the relationship that it bears to the human epidemiological study. Like any other phase of scientific investigation, it is the co-operation in different areas of scientific activity which furthers the achievement of a solution to any given problem. While only the epidemiological study can give definite proof of the relationship of smoking and lung cancer, the studies in the laboratory are essential in providing a practical solution to this problem, short of abolishing the smoking habit. Knowing that man will continue to smoke regardless of the evidence, we must expand our laboratory work in order to provide a practical solution to the problem. The thousands of e

1 322 FEB. 7, 1959 CONTRIBUTIONS TO TOBACCO--CANCER PROBLEM lives lost in every country each year from cancer of the respiratory tract demands that we expedite our efforts. It is hoped that with the evidence already at hand a practical solution may be within our reach. It is toward this end that laboratory studies involving the smoking-cancer problem must now be directed. ' REFERENCES ' Auerbach, 0., Gere, J. B., Forman, J. B.. Petrick, T. G., Smolin, H. J., Muehsam, G. E., Kassouny, D. Y., and Stout, A. P., New Ertgl. J. Med., 1957, 256, 97. ' Badger, G. M., Buttery, R. G., Kimber, R. W. L., Lewis, G. E., Moritz, A. G., and Napier, I. M., J. chent. Soc. In press. ' Van Duren, B. L., and Nelson, N., Proc. Amer. Ass. Cancer Res., 1958, 2, 353. ' Blacklock, J. W. S., Brit. J. Cancer, 1957, 11, 181. ' Bock, F. G., Proc. Amer. Ass. Cancer Res., 1958, 2, 282. ` Bonnet, J., and Neukomm, S., Heiv. chint. Acta, 1956, 39, 1724. r Campbell, J. A., Brit. J. exp. Path., 1936, 17, 146. 4 - Brit. med. J., 1942, 1, 217. • Campbell, J. M., and Lindsey, A. J., Brit. J. Cancer. 1957, 11, 192. '" - - ibid., 1956, 10, 649. 't Cardon, S. Z., Alvord, E. T., Rand, H. J., and Hitchcock, R., ibid., 1956, 10,485. " Chang, S. C., Cancer, 1957, 10, 1246. " Cooper, R. L., and Lindsey, A. J., Brit. J. Cancer. 1955, 9, 304. t' Croninger, A. B., Graham, E. A., and Wynder, E. L., Cancer Res. In press. " Engelbreth-Holm, J., and Ahlmann, J., Acta path. rnicrobio(. scand,, 1957, 41, 267. t.` Essenberg, J. M., Horowitz, M., and Gaffney, E., West J. Surg., 1955, 63, 265. t' Gilbert, J. A. S., and Lindsey, A. J., Brit. J. Cancer, 1956, 10, 642. " - - ibid., 1956, 10, 646. " - - ibid., 1957, 1l, 398. 30 Graham, E. A., Croninger, A. B., and Wynder, E. L., Cancer Res., 1957, 17, 1058. 't Hamer, D., and Woodhouse, D. L., Brit. J. Cancer. 1956, 10, 49. " Hamilton, J. D., Sepp, A., Brown, T. C., and Macdonald, F. W., Canad. med. Ass. J., 1957, 77, 177. " Hoffmann, D., and Wynder, E. L., " A Study of Tobacco Carcinogenesis. VII. The Higher Aromatic Polycyclics." In preparation. '* Kuschner, M., personal communication. " Lacassagne, A., Zajdela, F., Buu-Hoi, N. P., and Chalvet, H., C.R. Acad. Sci. (Paris), 1957, 244. 273. " Lam, J., Acta path. nticrobiol. scand., 1955, 36, 503. 'T - ibid., 1955, 37, 421. " - ibid., 1957, 40, 369. _' Latarjet, R., Cusin, J: L., Hubert-Habart, M., Muel, B., and Royer, R., Bull. Ass. Jrane. Cancer, 1956, 43, 180. . " ° Leard, H., Jahn, A., and Hausbeck, C., Angew. Chem., 1956, 68, 212. " Leuchtenberger, C., Leuchtenberger, R., and Doolin, P. F., Cancer, 1958, 11, 490. " Lorenz, E., Stewart, H. L., Daniel, J. H., and Nelson, C. V., Cancer Res., 1943, 3, 123. " Lyons, M. J., Nature (Lond.), 1956, 177, 630. '• - and Johnston, H., Brit. J. Cancer, 1957, 11, 554. 's Magnus, H. A., J. Path. Bact., 1939, 49, 21. " Miller, L. M., and Monahan, J., Reader's Digest, July, 1958. 'r Neukomm, S., Oncologia (Basel), 1957, 10, 137. '" Orris, L., Van Duuren, B. L., Kosak, A. I., Nelson, N., and 39 .n .1 42 43 44 45 .e 4; 43 49 s9 sx 53 s. 34 , Schmitt, F L., J. nat. Cancer Inst., in press. Passey, R. D., Roe, E. M. F., Middleton, F. C., Bergel, F., Everett, J. L., Lewis, G. E., Martin, J. B., Boyland, E., and Sims, P., A. R. Brit. Emp. Cancer Campgn., 1954. 32, 60. Rockey, E. E., Kuschner, M., Kosak, A. I., and Mayer, E., Cancer, 1958, 11, 466. Roffo, A. H., Dtsch. med. Wschr., 1937, 63, 1267. - KrebsJorsch, 1939, 49, 588. Seelkopf, C., Z. Lebensmitt- Untersuch., 1955, 100, 218. Sugiura, K., Gann, 1956, 47, 243. Touey, G. P., and Mumpower, R. C. II, Tobacco Science. 1957, 1, 33. Wynder, E. L., Graham, E. A., and Croninger, A. B., Cancer Res., 1953, 13, 855. - - - ibid., 1955, 15, 445. - Lupberger, A., and Grener, C., Brit. J. Cancer, 1956, 10, 507. - and Wright, G., Cancer, 1957, 10, 255. - Kopf, P., and Ziegler, H., ibid., 1957, 10, 1193. - and Mann, J., ibid., 1957, 10, 1201. - Gottlieb,. S., and Wright, G., ibid., 1957, 10, 1206. - Wright, G., and Lam, J., ibid., in press. - - - in preparation. -- unpublished data. Bamstt MEDICAL JOURNAL LUNG CANCER MORTALITY AND THE LENGTH OF CIGARETTE ENDS AN INTERNATIONAL COMPARISON BY R. DOLL, M.D., D.Sc., F.R.C.P. A. BRADFORD HILL, C.B.E., F.R.S. Statistical Research Unit of the Medical Research Council P. G. GRAY, B.Sc. AND E. A. PARR, B.A. Social Survey Division of the Central Office of Informatiorl Study.of the recorded death rates from lung cancer in different parts of the world shows that there is a fairly close relationship between the present national mortality from the disease• and the national consumption of cigarettes 20 to 25 years ago. Data for 16 countries are given in Table I and illustrated in the Chart, in which for each country the standardized mortality of men in 1952-4 is set against the consumption of cigarettes per adult (of both sexes) in 1930. The latter TABLE I.-Mortality from Lung Cancer and the Consumptiort oJ Cigarettes in 16 Countries' • r Standardized Mortality of Men from Cancer of Lung in Years 1952-4. Rate per Million Cigarette Consumption i Y Mean Cigarette t i C Mortality I Country j Group I Rate n ear 1930 (per adult) onsumpt o.. (Unweighted) Over 300 England and Wales 461 1,378 l ~ Fin land 433 1,662 } 1 1,330 Austria 380 960 _ J_I 200-299 Netherlands .. 276 I 632 Belgium 254; 1,066 I Switzerland .. 236t 706 }j 840 New Zealand 216 478 U.S.A. 202t 1,296 J 100-199 Denmark 179 465 Australia 177 504 Canada 176 l 550 France 140 585§ Italy .. 1lOt 455 J Under 100 Sweden 89t 388 Norway 77 359 ~ 490 Japan 40 723 ' The standardized mortality rates were calculated by Segi (1957). Rate for cigarette consumption were given by Todd (1957) or were derived from data for Finland and Norway published by Nielsen and Clemmesen (1954), for Switzerland published by Gsell (1951), and for New Zealand kindly provided by the New Zealand Government, Department of Statistics. t 1951-3. j 1954. § 1951. ~ Soo u Z p Jz 400 7) W °2 Il _ a 40f W W OO k 300 ~ io a cgc 0 250 500 750 1,000 1,2SO I,SOO 1,750 NUMBER OF CIGARETTES SMOKED PER ADULT (BOTH SEXES) PER YEAR (1970) Relationship between lung cancer mortality and previous cigarette consumption in 16 countries: The regression ltne is given by y=0.24x+28; the correlation coefficient is 0.76. t t

I w 11 • OF FEB. 7, 1959 LUNG CANCER AND LENGTH OF CIGARETTE ENDS • date was chosen to allow for a reasonably long induction period for the lung cancer ; it is not possible to test the effects of lengthening this period still more, since there is insufficient information about cigarette consumption in previous years. For the same reason it is not possible to use the cigarette consumption of males only, but as long ago as 1930 the consumption by women was probably only a small part of the total. In addition to these defects in the data we must note the great hazards of all such international comparisons. Firstly, the standard of death certification is likely to vary from one country to another, so that the numbers of deaths attributed to lung cancer in 1952-4 will not necessarily be a true measure of its occurrence in those years, and the degree of error will doubtless vary. Secondly, the average consumption of cigarettes per adult in 1930 is likely to be an inadequate measure of the extent to which a population has been exposed to risk from smoking. For example, (t) the spread of cigarette smoking is unlikely to have been the same in all countries : it may well have affected different sex and age groups at different periods ;(2) hand-rolled cigarettes are excluded from the estimates of the number of cigarettes smoked, and the frequency of their use will vary from country to country ;(3) no allowance is made for any possible carcinogenic effect associated with the smoking of tobacco in cigars or in pipes ; and (4) there may be substantial differences in the methods of smoking, in the kinds of tobacco used, and in the quality of the smoke reaching the bronchial mucosa. Moreover, even if exact information was available about all these factors, it would still be difficult to be sure of making a valid comparison between consumption and mortality, since so little is known about the mechanism of carcinogenesis and the relationship between the duration and intensity of exposure and the incidence of the resulting disease. In spite of these various limitations to the data there is, nevertheless, a distinct relationship in these 16 countries between their cigarette smoking of 1930 and their lung cancer death rates of 1952-4. The principal exceptions are the U.S.A. and Japan, in both of which the mortality is substantially lower than would be expected from their apparent level of cigarette consumption (Table I). For the U.S.A. it has, however, often been suggested that the habit of throwing away a large unsmoked butt may contribute to the relatively low mortality. This possibility was particularly stressed by Sir Ernest Kennaway (1957), who made some preliminary measurements in this country. Factual data for the size of discarded butts on a large scale have, however; been lacking until the recent report by Hammond (1958). Cigarette Ends U.S.A. In Hammond's study of the habits of the U.S.A. 4,283 butts were collected from ashtrays in homes, offices, and restaurants or picked up from sidewalks, stations, and parks in four large cities and several smaller cities and towns scattered throughout the country. The average length of all the butts was 30.9 mm., or some 40% of the length of the average cigarette. It did not differ appreciably between cigarettes with filter-tips (31.0 mm.) and those without (30.7 mm.), nor did it vary greatly between the butts collected in the different localities, save that it was somewhat longer in restaurants (34.5 mm.) and shorter in parks (26.7 mm.). eRntsx 373 MEDICAL JJURNAL It seems likely, therefore, that the average figure for the whole series may be a reasonable estimate of the true average for butts found by these means in urban areas of the U.S.A. On the other hand, since the butts were collected some time after they had been thrown away it is likely that sometimes a substantial amount of the cigarette will have burnt af ter the cigarette was discarded (a factor which might contribute to the relatively low average length of butts picked up in the open in parks). The average given by these data is therefore likely to be an underestimate of the length of the butt at the time it was actually put out or thrown away by the smoker. ` England and Wales Corresponding data for the length of the butt discarded in Britain have now been obtained in the study here reported, which was carried out by the Social Survey Division of the Central Office of Information (Gray and Parr, in the press). In this study a different technique was adopted. A letter was sent to a small randomly selected sample of the civilian population of England and Wales, aged 21 years and over (picked from the 1958-9 Electoral Register by choosing the first number under 150,000 at random and then taking every successive 150,000th electoral number). The recipient was asked in reply to give particulars of his (or her) usual smoking habits and, if a cigarette smoker, to collect, in a tin enclosed with the letter, the butts of all the cigarettes smoked the day after the letter was received. To avoid any change of habits the specific object of the inquiry was not mentioned or even hinted, the wording of the request being as follows: " We are carrying out an inquiry for the Medical Research Council. As you may know, they are studying the effects of smoking on people's health. They want to collect for examination a sample of cigarette ends which have just been smoked in the normal way. You have been picked purely by chance for this inquiry. Please will you help us ?" It is a tribute to the co-operativeness of the British public that, of the 200 thus written to, 180 (90%) responded (with the same proportion in each sex). Of the 81 men who replied, 38 said that they smoked cigarettes only and 4 that they smoked cigarettes and a pipe ; 6 said that they smoked a pipe only and 33 that they were not then smoking. Of the 99 women who replied, 29 said that they smoked cigarettes and 70 that they were not smoking. The proportion of •current smokers is unexpectedly small. This may have been due partly to a tendency for smokers to have replied less readily than non-smokers; but a few irregular smokers may have regarded themselves as non-smokers for the sake of simplicity. The butts returned by the cigarette smokers amounted to approximately 80°0 of the declared number usually smoked. They were straightened out and their length was measured. Mean values for different groups of interest are given in Table II. It will be seen that the average length for all 772 butts was 18.7 mm.,* or a little over a quarter of the ordinary 70-mm. cigarette. The various subdivisions of the table are more remarkable for their similarity than dissimilarity, and the only formally significant difference is between men and women, where the latter show a slightly shorter mean length. This is due in part to the *The sampling standard error allowing for the within-persorw correlation is 0.4 mm.

324 FEB. 7, 1959 TABLE 11 -Mean Length oj Cigarette Ends Provided by a Samp/e o/ Adult Population of England and Wales No. of No of Average Group Smokers . Butts Length of Sending in Butt in Butts in mm. AII cigarette smokers 71 772 18•7 Sex Male { 42 540 19-3 Female .. .. 29 232 l7•3 Age of r21-35 .. 12 184 19-8 males { 36-55 .. 17 158 20.0 :: l56 and over 13 198 18-3 Usual No. of r 1-4 8 16 19•1 5-14 .. cigarettes J 28 228 18•9 5 smoked 15-24 .. 30 428 I8•5 ` per day 25+ .. .. 5 100 l9•0 Type of J Filter-tip .. 11 • 99 l7•6 cigarette lOrdinary tip .. 66• 673 18•9 * 6 persons sent in both. fact that 27%, of the butts returned by women were filter-tipped, against only 71/%' of those returned by men. There is no trend apparent either with age or with the usual amount of smoking. (If the usual amount for each individual be used to recalculate the overall average a figure of, 18.5 mm. is reached, which differs negligibly from the 18.7 mm. for the butts actually returned. But we do not, of course, know whether the missing butts were, in fact, different from those sent in.) This lack of difference between the subgroups allows us with more confidence to contrast our total figures with those from the U.S.A., where such subdivisions of the data are, owing to the method of collection, impossible. LUNG CANCER AND LENGTH OF CIGARETTE ENDS England and Wales Compared with the U.S.A. The frequency distributions of the lengths of butts measured in the British and the U.S. inquiries are contrasted in Table Ill. The differences are very large. TABLE III.-Comparison o/ Cigarette Ends Measured in Englar:d and Wales and the U.S.A. Length in mm. England and Wales (%) U.S.A. (%) Up to 9-5 t0-14•5 15-19•5 20-24•5 25-19•5 30- 3'•5 35-3)•5 40 r .. Total No. of butts measured Average length (mm.) .. 1-2 17-5 46•7 25-4 6•9 1•4 0-6 0-3 772 (100%) 18•7 0•8 7•9 19•7 19•9 20-9 12-8 18-0 } 4,283 (100%) 30-9 In total the average length of the U.S. butts (30.9 mm.) is 65% greater than the British average (18.7 mm.). More striking still is the fact that 2 out of every 3 British butts (65.4%) were less than 20 mm. in length and less than I in 10 of the American were smoked down to this level (8.7°~)). At the other end of the scale only 1 in approximately 40 of the British butts were 30 mm. or longer (2.3%), while as many as half the American ends were discarded at such lengths (51.7%). Though in both countries the filter-tipped butts did not differ greatly from the ordinary tipped, we may note that only 13 %, of the English butts were filter-tipped, compared with 60%, in the American sample. Discussion The statistics on cigarette ends reported here for this country are admittedly on a small scale. Those obtained by Hammond previously for the U.S.A. are based upon larger numbers but form a less clear sample Bttrtisx MEDICAL JOURNAL ~ of the whole population. The great contrast between them. however, strongly supports the general impressions of any careful observer of the two countries ; there can be no doubt that in substance it is true. Such a very great difference in the British and American habits at the present time makes it at least possible that the greater butt length in the U.S.A. can account for much of the relative deficiency in its lung cancer mortality. If carcinogenic substances are associated with the polycyclic hydrocarbons in cigarette smoke, it must be anticipated that they would be produced by combustion, condensed further along the cigarette and redistilled as the cigarette burns shorter (Gilbert and Lindsey, 1957). Quite a small difference in butt length might therefore result in a substantial difference in the amount of carcinogen inspired. On the other hand, we must note that the present data may not be a good indication of the relative lengths of the butts which were discarded in the U.S.A. and Britain 20 to 30 years ago. Substantial changes are known to have taken place in smoking habits in the U.S.A. in the last few years ; the proportion of filter-tipped cigarettes has increased greatly and the proportion of extra large cigarettes has also risen-that is, cigarettes 80 to 85 mm. long against the standard 70 mm. We know of no evidence to show whether or not the average length of butt has altered ; personal impressions suggest that the difference between the two countries is of long standing. We do not suggest that this difference in butt length is the only factor that might account for the relatively low mortality from lung cancer in the U.S.A. Many other factors may have contributed to it, and, in so far as they concern the difference between the mortality in England and Wales and in the U.S.A., they have been discussed in detail by Hammond (1958). We are concerned to emphasize that such an exception to the broad geographical association between present-day lung cancer mortality and past national smoking habits, as shown in Table I, may have many explanations. It cannot be lightly taken at its face value and accepted as wholly incompatible with the general evidence on cigarette smoking. The only other extensive data on cigarette ends of which we are aware were obtained by Korteweg for Holland (referred to by Hammond, 1958). These gave an average length of 19.7 mm. for 545 butts, a figure akin to the one we have found for England and Wales. Summary In spite of the well-known difficulties of international statistical comparisons there is a distinct relationship in 16 countries between their reported cigarette consumption in 1930 and their lung cancer death rate in 1952-4. Striking exceptions are Japan and the U.S.A., which had a relatively high cigarette consumption but. a relatively low incidence of cancer of the lung. In the U.S.A. one explanation may lie in the larger cigarette end that the citizen discards. A comparison of cigarette ends, specially collected and measured in Britain, with similar data recently reported from the U.S.A., gives an average length of 30.9 mm. in the latter country against 18.7 mm. in the former. Two out of three cigarette ends in the British sample were less than 20 mm. in length ; only I in 40 was as long as 30 mm. In the American sample less than I in 10 were smoked down to 20 mm. and half were discarded with 30 mm. or more still unsmoked. Such a large national difference in a r y •

FEB. 7, 1959 LUNG CANCER AND LENGTH OF CIGARETTE ENDS BRITISH 325 MEDICAL JOUANAL . , v P habits might considerably influence the national exposure to any carcinogen in cigarette smoke. We are most grateful to Mr. Louis Moss, Director of the Social Survey, for his advice and for the facilities which enabled the British data to be collected. . REFERENCES Gilbert, J. A. S., and Lindsey, A. J. (1957). Brit. J. Cancer, 11, 398. Gsell, O. (1951). Schweiz. med. Wschr., 81, 662. Hammond, E. C. (1958). Brit. med. J., 2, 649. Kennaway, E. (1957). Ibid., 1, 299. Nielsen, A., and Clemmesen, J. (1954). Dai, med. Bull., 1, 194. Segi, M. (1957). " Cancer Mortality Statistics in Japan." Department of Public Health, Tohoku University, Japan. Todd, G. F. (1957). Statistics of Smoking. Tobacco Manufacturers' Standing Committee, Research Papers, No. 1. TOBACCO SMOKING, RESPIRATORY SYMPTOMS, AND VENTILATORY CAPACITY STUDIES IN RANDOM SAMPLES OF THE POPULATION BY I. T. T. HIGGINS, M.D., M.R.C.P. Member of Scientific Staff, Pneumoconiosis Research Unit of the Medical Researcli Council, Llandough Hospital, Penarth, Glamorgan Evidence is accumulating that tobacco smoking is an important factor in the aetiology of chronic bronchitis and emphysema. In a prospective study of mortality of medical practitioners Doll and Hill (1956) found a statistically significant gradient from non-smokers to heavy smokers in the mortality rates due to chronic bronchitis. Investigations of bronchitis morbidity in hospital patients, working groups, and representative samples of the general community support the hypothesis. Abbott et al. (1953) presented their findings in 294 patients suffering from various types of emphysema. They noted an almost constant history of protracted cough or so-called " chronic bronchitis " in these cases and considered that tobacco irritability was an important aetiological factor in 214. Palmer (1954), in a study of 422 male hospital patients, found that chronic bronchitis ~was significantly commoner in smokers than in non-smokers. Greene and Berkowitz (1954) recorded a very high prevalence of "smokers' bronchitis" in patients smoking 20 cigarettes or more daily. These findings were supported by Leese (1956), who compared 100 patients with bronchitis with 100 controls and showed that the controls smoked less. Lowell et al. (1956) studied 34 patients (28 males and 6 females) aged 50-81 years and concluded that in New England smoking was the major cause of eihphysema in this age group. Oswald and Medvei (1955) investigated the recordi of 6,245 clerical Civil Service employees and compared smokers with non- smokers. They found a higher prevalence of bronchitis in both sexes among the smokers. Clifton (1956) compared respiratory symptoms in a group of workers in heavy industry in Sheffield with those recorded in males in a rural general practice and noted a close association between tobacco smoking and persistent cough and sputum. Phillips et al. (1956) observed a close correlation between cigarette smoking and chronic cough, and C stressed the importance of standardizing for smoking habits in assessing the other factors considered important in the aetiology of chronic bronchitis. Ogilvie and Newell (1957) carried out an investigation of bronchitis in a sample of 3,866 inhabitants of Newcastle upon Tyne. They compared 464 bronchitics with 485 non- bronchitic controls and showed that in both sexes the bronchitics were th: heavier smokers, and among the men included a higher proportion of cigarette smokers. There was, however, no significant association between the number of cigarettes smoked and the disease. In a recent investigation of 1,062 men aged 60--69 living in Birmingham (Brown et a1., 1957) chronic bronchitis was found to be more frequent in cigarette smokers than in non-smokers. Our own investigations at the Pneumoconiosis Research Unit of random samples of geographically defined communities have supported these findings. Thus in men aged 55-64 studied in Leigh (Lanes) tobacco smokers had significantly more " chronic bronchitis," defined as persistent sputum and a history of one bronchitic chest illness during the past three years, than non-smokers (Higgins et al., 1956). In the Vale of Glamorgan, in South Wales, the prevalence of persistent cough and sputum was significantly higher in smokers than non-smokers, and smokers also appeared more liable to chest illness, breathlessness, and " chronic bronchitis " as previously defined, though for none of these symptoms did the difference reach the conventional level of statistical significance (Higgins, 1957). Finally, in Annandale, in South-west Scotland, a similar trend was observed in a sample of men and women aged 55-64 (Higgins and Cochran, 1958). Surprisingly few attempts have been made to correlate smoking with changes in lung function. Whitfield et al. (1951) studied the effect of tobacco smoking on lung volume in 58 healthy men, and concluded that smoking produced a slight diminution in the vital capacity and a more pronounced increase in the ratio of residual air to total lung volume. These conclusions were criticized by Oldham (1951), who pointed out that they could not be drawn with certainty from the results presented. Bickerman and Barach (1954) studied the acute effects of smoking three cigarettes in 91 patients with emphysema and 27 normal subjects. Although they observed no consistent trend, they noted a fall in the maximum breathing capacity in 10 subjects and suggested that this might have indicated increased bronchospasm. More recently Eich et al. (1957) investigated the acute effects of smoking on the mechanics of respiration in chronic obstructive emphysema. They showed an immediate statistically significant increase in the airway's resistance in 14 out of 15 emphysematous patients after smoking one cigarette, but no change in a control group. Franklin (1958) has reported the results of an investigation of the effect of smoking habits on lung function in a working population aged 40-65. Analysing kymographic tracings of the maximal forced expiration, he found that the flow rate was significantly lower in heavy smokers compared with light smokers or non-smokers. There is clearly a need for further studies assessing the long- term effect of smoking habits on lung function, using an objective physiological test. Our investigations of bronchitis in different communities have always included detailed smoking histories and also an assessment of the ventilatory capacity. The purpose of this paper is to consider the relation of smoking to

326 FEB. 7, 1959 TOBACCO SMOKING AND RESPIRATORY SYMPTOMS symptoms and to the ventilatory capacity as assessed by the indirect maximum breathing capacity test. Methods and Procedure Assessment of Ventilatory Capacity.-We record the three-quarters-second forced expiratory volume (F.E.V.), using a Gaensler type spirometer with electronic timer. The first measurement is rejected and the mean of three subsequent ones is multiplied by 40 and expressed as the indirect maximum breathing capacity (M.B.C.). • The validation of this test as an objective measure of respiratory disability and the reasons for preferring it to the M.B.C. measured directly have been considered previously (Carpenter et al., 1956). Its accuracy, simplicity, and repeatability are characteristics admirably suited to field investigations, particularly when these are continued over a number of years. Smoking Habits.-In each investigation a questionary has been used to record respiratory symptoms and smoking habits. Present smoking habits and major changes of habit over the past 10 to 15 years have been noted. To reduce the error due to observer variability to a minimum (Cochrane et al., 1951) all the interviews were carried out by one observer. The symptoms have always been recorded before asking about smoking habits. Smokers have been classified after the manner suggested by Doll and Hill (1950) according to the quantity of tobacco smoked. A smoker is defined as one who at any period of his life has smoked 1 g. of tobacco or more a day for one year. One cigarette is equivalent to 1 g. of tobacco a day ; 1 oz. (28 g.) of tobacco a week is equivalent to 4 g. of tobacco daily. The subjects were classified into non-smokers, ex- smokers, and smokers of 1-4, 5-14, 15-24, 25-49, and 50 g. and over a day. For most purposes a simple classification of smokers into light (1-14 g.) and heavy (15 g. a day and over) was used. The classification was based on the stated average consumption over the previous 10 years. The separate classification of ex- smokers explains certain discrepancies between the BRR7SA MEDICAL JOURNAL t figures in this paper and those published previously (Higgins et al., 1956) in which ex-smokers were included in their appropriate group when their tobacco consumption was averaged over 10 years. The Samples Stttdied.-Detailed accounts of the methods used to obtain our random samples have been published previously (Thomas et al., 1956 ; Higgins et al., 1956 ; Higgins, 1957). The samples were drawn from the Rhondda Fach and Vale of Glamorgan, in South Wales, Leigh in Lancashire (England), and Annandale in South-west Scotland. The Vale of Glamorgan and Annandale are rural areas and the Rhondda Fgch and Leigh are urban areas. In the Rhondda two samples were studied. The first was stratified by age, occupation, and, in the case of the miners and ex-miners, x-ray category of simple pneumoconiosis ; the second was drawn from all miners and ex-miners aged 55-64 living in the area, irrespective of x-ray category. In Leigh, again, a random sample of men aged 55-64, including non-miners, miners, and ex-miners with and without pneumoconiosis, was seen. In both the Vale of Glamorgan and in Annandale the samples considered here consisted of non-miners. Over 90 / of those drawn in the various samples were seen. Results Table I shows the numbers and mean indirect M.B.C., according to their smoking habits, in each age and occupation group. The miners and ex-miners seen in Leigh and the Rhondda have been subdivided on the basis of their chest x-ray category into those with and those without pneumoconiosis. In the second Rhondda sample and in Leigh there were a few miners with progressive massive fibrosis (P.M.F.), and these too are shown separately. Smoking Habits The proportion of non-smokers in the 55~4 age groups was between 5 and 10% of the samples. The proportion of non-smokers was somewhat higher in the TABLE I.-Mean Indirect M.B.C. According to Age, Occupation, and Smoking Habits in 734 Men seen in Five Random Samples. Percentages in Parentheses Smoking Habits (g.(Tobacco'Day) Area Age Group Occupation and X-ray Category Non-smokers 1-14 g.!Day IS g.!Day and Over Ex-smokers Totat No. . No. M.B.C. No. M.B.C. No. M.B.C. No. M.B.C. Rhondda Fach 25-34 1. Non-miners 3 (13-6) 145 7 9(40 9) 1342 9(409) 135.6 1(4-5) 1530 22 (99-9) (tst sample) 2. Miners and ex-miners cate- - gory 0.. 2 (4-3) 1430 20(42 6) 121.0 17 (36-2) 123.9 8(17-0) 132•1 47 (100-1) 3. Miners and ex-miners category 3 12 (25•5) 114•1 26 (55-3) 1234 9(19.1) 129 3 0 - 47 (99-9) 35-54 4. Non-miners 5(14-7) 1304 16 (47-1) 95.2 10 (29•4) 115-5 3 (8-8) 98-3 34(100.0) 55-64 5. Non-miners 3 (6-0) 93.0 21 (420) 88-9 17 (34•0) 98 6 9(18•0) 91•4 50(1000) 6. Miners and ex-miners category 0 2 (4-2) 82•0 28 (58-3) 71.8 14 (29-2) 57.0 4 (8-3) 75.3 48(1000) 7. Miners and ex-miners category 3 . 4(8•7) 93.3 27 (58-2) 77-8 11(23 9) 622 4(8•7) 97-0 46(1000) (2nd sample) 55-64 8. Miners and ex-miners category 0 .. 0 - 8(42-1) 746 10(526) 749 1 (5-3) 81•0 19(1000) 9. Miners and ex-miners with simple pneumoconiosis 1(5-9) 98•0 10(58 8) 71.5 6(35.3) 80.1 0 - 17(1000) 10. Miners and ex-miners with P.M.F. .. .. .. 0 ~ 7(63 6) 64-3 3(27-3) 71-0 1(9•1) 36•0 l1 (100-0) Leigh 55-64 11. Non-miners 8(9 5) 92-0 39 (46-4) 87-4 29 (34-5) 630 8(9•5) 71 9 84 (99•9) 12. Miners and ex-mine-s category 0 1 t(10 4) 80.9 49(48 5) 76-0 36 (35-6) 73.0 5 (5-0) 86•4 101 (1000) 13. Miners andex-miners with simple pneumoconiosis 2(9 5) 90-5 10 (47-6) 79-2 7(33-3) 91•6 2(9-5) 87•5 21 (99 9) 14. Miners and ex-miners with P.M.F. .. .. .. 1(100) 84.0 8(800) 76.3 1(100) 42.0 0 - 10(1000) Vale of 55-64 15. Non-miners 5(5•8) 101-2 30 (34-9) 98.1 34 (39-5) 82 6 17 (198) 828 86 (100 0) Glamorgan Annandale 55-64 16. Non-miners 6(6 6) 114.3 37 (40 7) 89.1 35 (38-5) 95 7 13 (14 3) 99.9 91(100.1) , r .