Document v6RerXkLXBXoow5emZOJRYKL8

r &S 109917 BIO-MEDICAL RESEARCH DOCUMENT DESCRIPTION FORM 63 Duplicate in all cards:--> /V ; year as-1961- oooaaes 7> File number [Right justify [Numeric only] Author(s), as Last Name FS (No Punctuation) and coden for journal as JAMA preceeded by.one blank space 20 -21 JF Grjs-htu'"4 P&- 40 41 77 Sub-Index Code 60 '61 62 u 12 Title of Renort: end with space-hyphen-hyphen-space. Follow with Index Terns, separated from each other with -comma-space. Avoid other punctuation; do not abbreviate. 12 61 62 !/ .t / ? ?'/? i, j f* 1 -f J 21 - -- --- 22 23 24 Source (Journal, Vol., Number, Pages,Date) 12 ------ _* - f-`. / F ' - /-A 1 J' j : 61 6231 32 Brief Su-TMp-ry' 12 10 SUMMARY: 61 62 61 62 63 64 r \ High, wise and handsome Table I. Heights and chest circumferences (inches) of 5732 Scottish militiamen, data compiled by an army contractor and printed in 1817 12). Height Chest circumference To- 33 34 35 __3_6_S_J7 38 39 40 41 42 43 44 45 46 47 48 tal 64-65 66-67 68-69 70-71 72-73 Total 1 1 I 0 0 3 7 31 69 108'"-154 142 118 66 17 6 3 0 0 0 0,i 727 9 30 2 16 14 78 170 341 442 34 91 187 '\34l 7 31 62 117 337 231 124 34 436 367 292 126 153 209 148 102 12 3 I 0 jtf' 1815 70 13 3j ) 1981 40 16 J-T 0 0 897 0 0 I 9 7 20'-\ 38 62 65 45 43 18^ 7 1 I 317 19 81 189 409 753 1062 1082 935 646 313 168^50 18 3 I 5732 Table 2. Data from Table I reclassified and expressed in/ounded percents, together with round ed percents (in parentheses) from a bivariate normal distribution with correlation .45. The means and standard deviations of the bivariate npflnal distributiomn were fixed to be 39V inches Vand 2'/t inches (chests) and 67Vi and 2'lt incites (heights) (J). Chestfcciirrcumference tieigm 33-35 36-38 39-40 41-42 43-45 4&48 tal 64-65 66-67 68-69 70-71 72-73 Total 1(1) 0 1(1) ^30 ( 9) 0(0) Jr 5( 5) 0JPT _0(0) 2 ( 1) 0( 0) 2(2) 23(21) 5 ( 6) 14(14) 14(13) J( 5) K I) 39 (39) H 2) 6( 8) 11(11) 6( 7) 2 ( 1) 26(29) 0(0) 1(2) 4(3) 3(3) 2(1) 10(9) 0(0) \, 13 ( 15) 0(0) \32( 34) 0(0) a4( 32) 0(0) 16 ( 16) 0(0) 5 ( 3) 0(0) 100(100) r &S 109918 t. .1 4 i The MultiRac fraction collector LKB's new MultiRac'' fraction collector is a real space saver. On the opposite page you can see that reservoir, column, pump, monitor and recorder all mount on it easily. And you can keep adding decks to take as much more equipment as you want. The new LKB fraction collector is bright too. It will collect from microliters to liters, adjust fraction size automatically according to 0D, stop all flow as the head traverses, and on command will channel all void volume to waste. Its good looks go well beneath the the surface: solid state electronics, rugged materials of construction and a clear, bright, unambiguous LED display all make for an instrument that's safe, dependable and easy to use. Contact LKB today for full details. LKB Instruments Inc. 1`2`2'21 Parklawn Drive Rockville. MD 20862 301: 881-2510 Telex: 8-9682 MA-Jll Clfcis No. 361 on Readers' Service Card unheralded Gauss who constructed his data using "techniques now considered not to have been invented until at least half a century later. As statisticians we decr^both in adequate statistical reporting and^nappropriate statistical analyses. But bo shortcomings are too common in the lit erature of the past and present to permit them alone to serve as evidence of fraud or intentional deceit. Dorfman's article has served the worthy purpose of stimu lating discussion highlighting the in adequacies of Bun's descriptions. But^ using Dorfman's inappropriate statistical techniques to detect fraudulent "'"data would be to condemn a major portion, if not ail, of empirical science as fabrica tion. Donald B. Rubin Educational Testing Sendee, Princeton fSew Jersey 08541 Stephen M. Stigler Centerfor Advanced Study in the Behavioral Sciences, Stanford. California 94305 claw means on another may. with data like these, produce a remarkably close fit to a straight line, like that exhibited in Dorfman's fig ure 1 (p. 254). where his "fabrication equation" and the means are displayed. 2. Edinburgh Med. Surg. J. 13. 260 (1817). Table F was constructed by aggregating separate tables for 11 different regiments. The compiieriof the tables is only identified as "an army,contractor, a gentleman of great observation and singular accuracy." Quetelet used tbe~ches( measure ments in his Learn , . . suFla throat des prob- abilites (1846) and other books. 3. Thc^figures in parentheses were found from a bivanate normal distribution with chest mean 39'/i, chest,-Standard deviation 2'In, height mean 68, height standard deviation I')i. and cor relation coefficient,-45. The columns were con sidered as corresponding to classes (0, 35.5), <35.5, 38.51. (38.5. 405). (40.5. 42.5). (42.5. 45.5). (45.5, i). The row classes were taken to be (64. 66). (66. 68). (68. 70). (70. 72). (72. 74). Apparently no Scotsmen bclow?64 inches or above 74 inches were admitted to the militia, so the distribution was truncated at these-values and renormalized so that the sum of the proba bilities for the given cells was 1.0. The bivanate) normal probabilities were found from the Tablet of the Bivariate Normal Distribution Function and Belated Functions (Applied Mathematics Series No. 50. National Bureau of Standards. Washington. D.C.. 1959). All standardized ceil boundaries were rounded off before entenng in the table, to eliminate the need to interpolate. Means, standard deviations, and the correla tion coefficient were chosen as being stan dard fractions near sample estimates, and they may not produce the best possible fit. In cidentally. published tables are ill-suited to this purpose, and the required computations seemed laborious to us. References and Notes l. Among the other problems with Dorfman's sta tistical analyses are that (i) Dorfman's assump tion in his reply (p. 251) that "column totals are not changed by the weighting along rows'' does not follow from our proposed method of con structing the tables. It is m fact a tacit assump tion of the conclusion he wishes to prove: the algebra here is irrelevant, and the conclusion that "the column totals are determined by the row totals" (p. 254) is incorrect. Oil It is not un usual to obtain high correlations among class means of approximately multivariate normal data. For example, the correlation of the row means and the height midpoints of the Scottish data (our Table I) is .99604. not far short of the value .99867 Dorfman (p. 252) found for Burt's means. Analogously, regression of one group of n9n - Chemical Carcinogens: Estimating the Risk We would like to respond to the letter by Hooper, Harris, and Ames (16 Feb., p. 602). in which the authors comment on what they feel are ". . . several er rors of fact and interpretation" in an ear lier series of articles on chemical carcin- SCIENCE. VOL. 205 ogens by Thomas H. Maugh II (Research News, 29 Sept. 1978. p. 1200; 6 Oct. 1978, p. 37). In particular, they express concern that one of us (P.J.G.) incorrect ly inferred that a threshold exposure to vinyl chloride existed at 50 to 150 parts per million (ppm) and consequently might have underestimated the risk from ex posure to low levels of vinyl chloride by ", . . more than a millionfold. . . ." In fact, the risk estimate alluded to (/) (which predicts a cancer risk of 10-> in workers occupationally exposed to I ppm vinyl chloride for 35 years) is per formed without any assumption of a threshold. Instead this analysis incorpo' rates pharmacokinetic principles to pre- diet the rate of production of a reactive intermediate in vivo after exposure to vi nyl chloride. Although we believe it is possible that thresholds for chemical carcinogens ex ist, our research on vinyl chloride has not provided evidence showing such thresholds. Indeed, attempting to pro vide irrefutable evidence of absolute thresholds for chemical carcinogens is simply an exercise in futility. Because it is impossible to prove that any chemical is totally without risk, it is essential to move toward rational risk assessment based on the best available technology. Furthermore, it is important to check any type of risk estimate against real data whenever possible. In the case of vinyl chloride this is possible because a survey of almost 10,000 workers occupa tionally exposed to vinyl chloride was re cently conducted by Equitable Environ mental Health, Inc. (EEH) (2). Exposure of these workers (which occurred before current standards were established) was great--very likely more than 200 ppm and certainly far more than 1 ppm. If Hooper et al. are correct in projecting a risk of I0-'-1 to I0-1 from exposure to I ppm vinyl chloride, then this group of workers, exposed to much higher doses, should have experienced several hundred to several thousand cases of hemangio- sarcoma. However, the survey failed to show this--only five of the almost 10,000 exposed workers had developed this type of cancer. A key factor in the extrapolation of the results of animal studies to humans is the role played by metabolic activation. Vi nyl chloride is one of a class of chemicals in which the proximate carcinogen is generated from the parent molecule by #Wm" icrosomal oxidation (3). In general, is type of metabolic reaction occurs ^^rniore rapidly in small laboratory animals than in humans. Investigators in another laboratory have studied the metabolism ai of vinyl chloride in several species, in- 21 SEPTEMBER 1979 The first microprocessor-controlled fraction collector... H LKB's MultiRac " LKB's MultiRac fraction collector is unlike any other in versatility of function, reliability of operation and simplicity of use. Just push a few buttons and you can program the unit to collect in any size vessel from test tubes to carboys -- by time, drop or LKB's precise volume. The 250 msec switch-over between tubes minimizes dribbling and a "stop ' feature can eliminate it entirely. Microprocessor control conserves test tubes by diverting column void volume to a built-in waste container. Still more tubes are conserved by varying fraction sizes according to OD. During peak elution, small fractions are collected for optimal resolution... in the valleys, tubes are filled to capacity. You can start your run and leave for an early dinner, knowing that when you come back the next day, not a single tube will have been wasted. Non-tipping spring-loaded racks accommodate everything from 8 mm test tubes to 28 mm scintillation vials to funnels feeding carboys. Racks fit in a tray that is removed in one easy movement. Keeping wet operations below the circuitry and using tough, inert materials make the MultiRac safe and easy to use with all liquids, even aggressive solvents and radioactive solutions. Power interruptions are also tolerated -- a memory retains data up to 16 hours. And in the unlikely event that something should go wrong, a self-diagnosis function helps you locate the problem. By the way -- fraction collector, pump, column, monitor and recorder all occupy less than two square feet of bench space. [333 LKB Instruments Inc. 12221 Parkiawn Drive Rockville, MD 20852 301:881-2510 Telex: 8-9682 Circle No. 362 on Readers' Service Card j? v.un 1207 R&S 109919 r Javelin 'fightViewing Devices eluding humans, and have determined that it is about 4.3 times slower in hu mans than in rats (4). Consequently, one would expect that humans should be much less sensitive than rats to the carci nogenic activity of vinyl chloride. The live rates of DN A replication in these tissues, estimated by injection of 3H-la- beled thymidine 48 hours after a dose of chloroform (10). The demonstration of chloroform-induced tissue damage at tu morigenic doses is particularly notewor i> T j `j jring risk assessment of Gehring et al. (/), thy because chloroform has been tested based on this principle, is reasonably by several investigators in the bacterial photographs consistent with the results of the EEH mutagenicity test developed by Ames survey. Ten hemangiosarcomas are with apparently negative results (II). out of the dark. predicted from exposure to 200 ppm vinyl chloride and five were actually observed. However, the risk estimate of Hooper et al., which apparently does not consider this principle, clearly cannot be recon Thus we do not feel tissue damage can be legitimately dismissed as a factor in the carcinogenicity of chloroform. Fur thermore, Indications that chloroform probably lacks genotoxic activity raise ciled with the data in the EEH report. serious questions about the validity of Hooper et al. also comment on the carcinogenic risk estimates based on the role of pathological tissue damage in the "one-hit" model, since the one-hit mod cancer bioassays of chloroform. Many el was developed to deal with genetic investigators have reported that short- events. and intermediate-term exposure to There is certainly no question that the chloroform is toxic to kidney and liver test developed by Ames can give useful tissue in mice (5) and rats (6). However, information about the potential of vari in the long-term studies conducted by ous agents to interact with DNA. How the National Cancer Institute (7) and Roe ever. so many complex interactions are (8) the absence of reported pathological involved in the generation of a tumor in a damage is very likely misleading. These whole animal that it is essential to carry studies were designed to detect irrever out a complete evaluation of the effects sible toxicity, such as the induction of of chemicals in whole animals before t, tumors. In each case, the chloroform was drawing any conclusions about their car administered daily for a period of 78 to 80 cinogenicity. weeks followed by several months of ob Richard H. Reitz No infrared to taint studies. servation before the animals were killed. Consequently, any reversible tissue John F. Qimst^P Philip G. Wvtanabe More and more, physical and damage would probably not have been Perry J. Gehring social scientists, technical photog apparent in these studies. However, pro Toxicology Research Laboratory, j. raphers and others are turning to Javelin Night Viewing Devices (NVDs) for photographing and see ing in the dark. For those performing experiments, the elimination of infrared light subtracts one more variable in their research data. longed cell regeneration after chemical insult may be very significant in the in duction of tumors. In fact, in the first published study (9) to link chloroform exposure with tumor induction in mice, the authors noted that liver necrosis was Health and Environmental Sciences. U.S.A., 1803 Building, Dow Chemical U.S.A.. Midland. Michigun 48640 References 1 R&S 109920 Javelin NVDs are presently being used for emission or "smokestack" consistently produced by doses of chloroform that were tumorigenic. These I. P. J. Gehring. P. G. Watattabe, C. N. Park. Tax. icol. Appl. Pharmacol. 49, 15 (19791. I. Epidemioioykal Study of Vinyl Chloride Work research; studies of the nocturnal habits of mammals, reptiles and investigators were careful to administer a dose of chloroform 24 hours before er!, Final Report (prepared for the Manufactur ing Chemists Association, Washington, D.C.. by Equitable Environmental Health, Inc.. Rock* insects; and sleep patterns of humans. A major TV network exposed drug use of American soldiers in Germany. Another network verified Highway Patrol complaints of night time driver abuses. Whatever you're studying or necropsy so that both short- and long term effects could be observed. Recent studies in our own laboratory have been concerned with the induction of tissue damage and subsequent cellular regeneration after single oral doses of ville. Md., January 19781. 3. P. G. Watanabe. J. A. Zemple. D. G. Pegg. P. J. Gehring, Toxicol. Appl. Pharmacol. 44, 571 (19781. 4. A. Butcher, H. M. Bolt, J. Filser. H. W. Goer- gens, R. J. Loib. W. Bolt. Verb. Dtsch. Ges. Arbeitsmed. IS. Ill 119781. 3. M. K. Deringer. T. B. Dunn. w. E. Heston. Prtjr. Sue. Exp. Biol. Med. 83, 474 U9531'. P. Shubik and A. C. Ritchie. Science 117. 283 photographing--don't be kept in the dark. Let a Javelin NVD open your chloroform to male B6C3FI mice. We found that tissue damage could be de (1933). 6. T. R. Torkelson, F. Oyen, V. K. Rowe, Am, Inii. Hvy. Aij.'t J 37. 697 (1976V eyes. A range of models is available to fit on any camera-still.movie or TV. tected microscopically in both the liver and kidney after chloroform treatment. 7. Report an the Coretnoyenesis Sioassay of Chlueofurm (National Cancer institute. Bethesda. Md., I March 1976). For details, contact: JAVEL3Nsssp* In the liver, cellular degeneration and ne crosis were present after a dose of 240 milligrams (mg) of chloroform per kilo 8. F. J. C. Roe. A. K. Palmer. A. N. Worden. N. J. Van Abbe. J Environ. Pathol. Toxicol. 2. 799 (1979): A. K- Palmer, A. E. Street. F. J. C. Roe, A. N. Worden, N. J. Van Abbe. iW.. p. 821. 9. A. B. Eschenbrenner and E. Millet. J. Natl. ELECTRONICS gram (kg) body weight, hepatocellular swelling w as noted after 60 mg'kg. and Cancer Inst, S. 231 (1945). 10. R. H. Reitz. J. F. Quast. S. Pocotte. Toxicology Research Laboratory. Dow Chemicai Co., un SuOi'dU'v Of W4uf Kidai A Company, me. KJDOE 15 mg'kg apparently had no effect. In the published observations. 11. H. Uehleke. H. Grein. M. Kramer. T. Werner. 6357 Arizona Circle Los Angeles. CA 90045 kidney, necrosis was observed after both 240 and 60 mg'kg, but not after 15 mg/kg Xenobiottca 7, 393 (1977); V. F. Simmon, K, Kauhanen. R- G. Tardiff. paper presented at the 2nd International Meeting of the Environmental 3 Phone (213) 641-4490 Telex 69-8204 (10). This correlated well with the rela- Mutagen Society. Edinburgh. Scotland, 1977. Circle No. 329 on Readers' Service Card SCIENCE. VOL. 203