Document ykz6ezYJvLvd98ZB2vNZ7GJrD

TOXICOLOGYAND APPLIED PHARMACOLOGY 64,"393-404 (1982) - . . .. .. ,. _... . \ .. .*_ . .. .. . 2 . . - . ARTHUR SELLAKUMBAERRN, ARD D.GOLDSTEIN,' AND ROY E. ALBERT .-,-rr;iqf.< t-!.;: e-:!-... k.:.:. ,;._ .. 'I .. ,.-,.,. . Laboratory of Inhalation Toxicology and Carcinogenesis. Institute of Environmental Medicine, New York . .. IU..n. iversity Medical Center, 550 First , .. , Received November 10. 1981; accepted March 25. I982 . Cells :Are Increased by Ingested Ethanol. ,B SELWCUMAR, A. R., GOLDSTEIN,~B. D., A macol. 64,'3932404.F'or 13 weeks, groups' benzene by inhalation, 6 hr/day, 5 days/w 4 days/week The number and type of bl .. . were determined at regular intervals. Anemia and lymphocytopenia were observed in the .. peripheral blood of all benzene and benzene/ethanol-treated groups. These cytopenias, how- . . . ever, were more severe in the benzene/ethanol-treated mice. In addition, there was a transient increase of normoblasts in the peripheral blood of benzene/ethanol-treated mice which was .. ..I . . .. not observed in mice treated with benzene alone or in control mice. Groups exposed to benzene, or benzene/ethanol displayed .mo- r.>e.s.-e~v.e. r. *e... in mice exposed t reduced marrow and splenic cell o benzene/ethanol. Specifically, ulanti these &'but. redu th d e.d..u..c t i o n s cellularities were were ' _ -..,.- ' '" .**. .. . . . characterized &decreased numbers of I numbers,of granulocytes and normobl a tran.sient.increase in the numbers of the transient appearance of n o h o b l k wasnot ob&rved in mice treated with - o.,-' ti '''*c.' .nal c y dc c;., h r ome P-450 (Sei I - i m**4c,r;rrrr. +,',%%.; tz et I~ ' <a..l...,. ..1~9.7\..8.). Benzene is.thought':to be.converted to he- ?_..n.. -sg-p*r-olB-eu-le-6o,n;B&.',fcieelrPlr'aco?ut4en,r.t3Ss.2-w,, 6er3.e.P,,aeonrbdipt&h8ee8rdadlafybrslooomofd.,ethxmpreoaesrurmroew^ic,eFrapeneGdr . matotoxic1m. 'etabo..l.ites:<:via. mixed function I adwing+en;us tail was for Firherd oxidase (Snyder et id.;1967; Gonasun et aZ., blood determinations. Test and control animals weR 1973). It has been shown that treatment of bled on the same day and within 30 min to minimk rats with ethanol will increase the rate of benzene metabol&m-in .vitro (Sato et al., 1980, 1981). Studies were therefore under- - taken in this laboratory to examine the effects-of-Ggestdethanol on the hematotox- dXerences due to handling, circadian rhythms, etc. Red and white blood cell counts were determined on a Caulter Model ZB-I blood cell counter (Coulter Electronics, Inc., Hialeah, Fla.). Differential counts were determined manually by Wright-Giemsa-stained peripheral blood smears. Peripheral normoblasts were counted on the icity of inhaled benzene. We recently reported that ingestion of ethanol did increase the cytopenic effects of inhaled benzene on the peripheral blood of C57B1 mice (Snyder et al., 1981). A second study on another basis of 100 white cells. Following tail bleeding, mice were killed by cervical dislocation and the spleen and both femurs were dissected from each mouse. After the spleen was weighed, a section was removed to mite a cell suspension. The remaining section was retained for histopathology. Cells were freed from the splenic cap group of C57B1 mice was performed to assess the effects of ingested ethanol and inhaled benzene on recognizable blood cells in the marrow and spleen. sule by gently pressing the section against a glass Petri dish with a stainless steel spatula. Cells from the broken capsule were drawn into a preweighed heparinized mi- =crocapillary tube. Marrow cells were obtained by moving the epiphyseal tip of the femur. inserting a pro weighed, heparinized microcapillary tube into the h- METHODS men, and drawing a sample of cells (Fruhman and Gordon. 1955). Exposure conditions. Male, 6-week-old C57B1/6J mice (Jackson Laboratories, Bar Harbor, Maine) were quarantined and observed for anomalous behavior for 2 weeks before being randomly distributed into six groups of 18 mice each. Each group was assigned to one of the following combined inhalation and ingestion + + +treatments: air water, air 5% ethanol, air 15% +ethhol, 300 ppm kni%ne water, 300 ppm benzene + +5% ethanol, 300 ppm benzene 15% ethanol. Solu- tions of ethanol in water were prepared on a weight/ weight basis. Ethanol was provided for a 4-day period from Monday afternoon to Friday morning and fluid Microcapillary tubes containing cell samplr; from both organs were reweighed and washed into separate 1-ml samples of a hypotonic solution consisting of t h a parts heat-inactivated fetal calf serum to one part deio*ized water. Single cell suspensionsof both samplw Weft produced by gently impacting the cells against the bot. tom of the centrifugation tube with a fire-polish4 PaCteur pipet. A 2&pl aliquot of each sample Was Gtbdrawn to perform nucleated d l counts on the Codtcr cell counter. The remaining suspensions were incubated at rwm temperature for 10 min and gently pelletd by centrifugation. After reducing the supernatant VOlUmC samples were resuspended to make marrow and VIcea ,I ETHANOL INCREASES BENZENE TOXICITY 395 TABLE 1 COMPARISONOF PERCENTAGE BODY WEIGHT CHANGE It SD pep after mt exposure rcc 0 4 32 63 85 +Air sham % ethanol 0% 5% 15% 100 (21.6)"" 103 k 5 . 115 2 6 118 & 7 124 k 3 '100 (19:s)" 109 rt 5 120 5 10 134 k 11 136 -+ 9 100 (21.6)" 106 rt 5 110 f.6 117 k 7 126 -+ 6 +300 ppm benzene % ethanol 0% 5% 15% 100 (20.6)" 103 k 6 110 k 1 1 115 5 14 121 f 15 100 (19.7)" 102 It 4 107 k 7 110 .fr 6 108 f 3 *100 (20.8)" 99 7 104 k 8 91 + 4 87' "Numbers in parentheses are average initial weights in grams. b Only one animal remained in this group. mean. A 200-cell differentialwas performed on smears which were stained with 1% dimethoxybenzidenecoun- tered with Wright/Giemsa (LoBue et al,, 1963). Body weights, parhology. Body weightswere recorded prior to the start of exposures and following scheduled deaths at 4,32,63, and 88 days. A necropsy examination was performed on each animal. Tissues sectioned from these animals included lung, liver, Spleen, thymus, and bone marrow. All sections were fixed in formalin and stained with hematoxylin and eosin for histopathological examination. Lymph nodes were examinetj only for re[- rtive size. Srurisrics.All data shown in Figs. 1-10 are presented as mean k one standard error. Data were evaluated for differencesby the Student's t test. Because the group was small (3) at each data point, the Welch (1947) approximation was used to determine the numbers of degrees of freedom and the Student's t statistic. Except for dara concerning peripheral normoblasts, all statis- tical tests were two tailed with p = 0.05 chosen aS the kvel of significance. No statistical treatment was a p PEcd to the peripheral normoblast data because of a h e variance in the values for benzene/ethanol-treated a are presented as SE. RESULTS General The mean daily benzene concentrations 2 SD for 4, 32, 63, and 88 days of exposure were 302 2 8.4, 301 8.0, 300 8.7, and 300 -C 8.2, respectively. There were no apparent differences in the rates of body weight gain between groups exposed to ethanol alone or benzene alone (Table 1). The two groups treated with bene zene/ethanol, however, showed different rates of weight change than the other groups. Mice exposed to benzene/% ethanol showed a slower rate of weight gain than mice treated with either b alone. Mice treated with b ethanol showed a .TABLE 2 -I COMPARISON OF AVERAGE DAILY FLUID AND ETHANOL CONSUMPnON (24- SD) .;;.i'&4,-5.r.i:p_:.,_h*,_e.,r:.,:.a. lT.i'&A&; :&;&$:&-td;*-:..'&. :,d+; .: c. I n .t s..2w: ,e_r.e,dIepressed ,iri'-allg-roups trwted .,with -. . . .. I benzene ' (F' i.g.s.:.' 1':'and A,:,'-. 2). In general, -.-..peripheral erythrocyte counts were more de- ~...Cr0ir . , . .! :.*. I LA. pressed in benzen*e. /.e..:.tha<; no. l::*txr-.eated mice than .,.,,I.in*._m.:,.,ice.treate*d..'9w..'i.t'h;ben.z.en1 ep'al~ o:. ne, with the group ingesting'- 15% ethanol showing the ' most>:,. ... , ,.severe decline:' Peripheral lymphocyte ,counts 'were.severelydepressed in all groups exposed.to benzene. .At all times the lym- .p-hocytopenia in benzene/ethanol-treated mi& was greater than'or equal to that in mice t.r..e..a.t.e.,d. with benzene'alone.'" ice'.treated''wiith. e.tih_a. .nol alone showed hera1.'- erythrocyte and lymphocyte counts that were equivalent to control values. Peripheral neutrophil counts were statisti- cally equivalent among all six groups re- gardless of treatment. By the fifth.week of exposure, elevated levels .of normob.la..s..ts..,:appeared, in the pe- riphzral blood of the benzerielethanol-treated mice '(Fig. ' 3). No peripheral normoblasts were seen in the 'groups'exposed to benzene alone or ethanol alone. The appearance of peripheral normoblasts peaked for the ben- zene/l5% ethanol group and the benzene/ 5% ethanol group at approximately 32 and 63 days, respectively. '. Blood cell morphology of the benzene and benzene/ethanol-treated groups included the appearance of polychromatic and stippled mature red cells, atypical _.atelets, ,*:-i_. . increased cytoplasmic' vacuolization, neutrophilic hy- -tr.e-a-ted groups -andwere most severe in the benzene/ 15%-*ethanol-trea&d group. Macrocytosis and hypochromia were not ob served in any benzene-treated groups. B l d cell morphology was normal in groups treated with ethanol alone.. / -_, . . 4. Marrow and Spleen Cellularity Marrow cellularities of all benzene-exposed mice were initially depressed after 4 days of exposure and remained depressed at relatively constant levels for the 88 days of exposure (Fig. 4). Benzene/ethanol-treated mice showed more severe depressions of marrow cellularity than mice treated with benzene alone, and the severity of the depression was related to the dose of ethtnol. Spleen cellularities of all benzene-exposed mice were initially depressed after 32 days of exposure and remained depressed throughout the exposures (Fig. 5). Similar to marrow cellularities, spleen cellularities were more depressed in mice treated with benzene/ethanol than in mice treated with benzene alone. By the last exposure, the severity of the depression of spleen cellliIaritY was related to the dose of ethanol. Earrow and spleen cellularities were not different among air sham-exposed mice, regardless of ethanol treatment. Marrow normoblasts from all benzene-exposed groups declined in a fashion similar to total marrow cellularity (Fig. 6 ) . Depres- + +_.- - FIGS. 1-10. For all figures. points are mean values & SE. 0, Air water; 0,air 5% ethanol; + + + +. air 15%ethanol; e, benzene water; W, benzene 5%ethanol; A,benzene 15% ethanol. (a) Denote' statistically less than corresponding control; (b) denotes statistically less than benzene and water value>. (c) denotes statistically less than benzene and 5% ethanol values. ..- 1- ll'"b -I- A 4.0 0 I III 20 40 I 1' I I 60 80 Days on Study FIG. 1. Red blood cells. 16,000 T 42,000 - aI -\ -VI 8,000 uQ 4,000 I 100 0 398 BAARSON ET AL..`-:-- ." , - 250 100 50 01 0 . 20 40 60 80 Days on Study 100 FIG. 3. Peripheral normoblasts. ethan'ol-treat normoblasts from `benzene/ethanol-treata mice increased markedly after the first exposure week, peaked' between ` 32 and 63 days, and then declined (Fig. 7). This increase and decline of splenic normoblasts were essentially the same for both benzene/ ethanol-treated groups and coincided with the increase and decline of normoblasts in 1. the peripheral blood of these mice. A more :-- gradual increase in splenic normoblasts was seen in mice treated with benzene alone. This increase, however, was not associated with the appearance of any normoblasts in the peripheral blood. No differences were observed in marrow or splenic normoblast levels among any of the air sham-exposed mice regardless of ethanol treatment. Marrow and splenic lymphocytes were significantly depressed in benzene and ben- 2.8 T 0 II I1I I I I 0 20 40 60 a0 Days on Study FIG. 4. Marrow cellularity. ... 1 109 3.8 3.4 3.0 E" \ 2.6 9 -I -L .Yo 2.2 1.8 I.4 1.0 0 20 40 60 Days on Study FIG. 5. Spleen cellularity per milligram. I I __.. FIG. 7. Normoblasts (spleen). L I0.2- "n I 0 ,1 p,b I nh --?" II I I 20 40 60 80 100 Days on Study FIG. 8. Lymphocytes (marrow). 400 4.0 00 - I IIII1 I I 20 40 60 80 Days on Study FIG. 9. Lymphocytes per milligram (spleen). I1 100 .I \. : - .. 402 ' r.j -. < : . thology and Histop ..EFFECT OF ON S BENZENE IZE OF THYMUS ..-Lc .,l_," .-+ _. No significant\gross abnormalities were ,..* ,' apparent at autopsy in any organ examined *'A xr . J c- Day after.firset x ~ u r .e "-from air sham-exposedmice or A c e exposed .I -Group', e ~ _ _ -+Air water - d +Air 5%ethanol +Air 15% ethanol +Benzene water +Benzene 5% ethanol +Benzene 15% ethanol I. :4 . " ' 3 2 - 63 _- 88 to benzene alone. In mice treated with hen- zene and ethanol; the thymus was reduced N b N N 'N to 50%of original sizeor couldnot be readily N N N N detected by the 32nd'day of-kxposure. N N N N NN NN I. 1 AA N N .: 1- shown in Table 3, the incidence and extent of thymic involution were related to the dose . A of ethanol. By the 32nd day of exposure, the thymus was reduced to 50% of control size Three animals per group. 'N, normal;I, greater than 50%involuted; A, absent or unidentified. in benzene/5% ethanol-treated mice, but could only be detected microscopically in benzene/ 15% ethanol-treated mice. Microscopic examination of tissues re- zene/ethanol-exposed mice. The depression vealed few treatment-related abnormalities of marrow lymphocytes was similar to the other than hypoplasia in the marrows and depression in total marrow cellularity. Counts spleens of benzene-treated mice. Splenic and were initially depressed after 4 days of ex- marrow hypoplasia was more prominent in posure and remained depressed at constant benzene/ethanol-treated mice than in mice levels throughout the study (Fig. 8). Ben- treated with benzene alone. No evidence of zene/ethanol-exposed mice were more se- megaloblastoid marrow was observed in any verely affected than mice treated with ben- animal. Mice receiving benzene and ethanol zene alone. Splenic lymphocytes were de- did show some very minimal periportal vac- pressed relative to controls 32 days after the uolization in the liver. first exposures in benzene/ethanol-treated mice and 63 days after the first exposure in mice treated with benzene alone (Fig. 9). DISCUSSION The depressions of splenic lymphocytes were more severe in benzene/ethanol treated mice These results demonstrate that, in C57B1 than in mice treated with benzene alone. mice, the combination of inhaled benzene Marrow granulocytes were depressed af- and ingested ethanol is more hematotoxic ter 4 days of exposure in benzene/ethanol- than inhaled benzene alone. Compared to treated groups and after 32 days of exposure treatment with benzene alone, the combined in mice treated with benzene alone (Fig. 10). treatments produced a more marked depres- The severity of cellular depression on ben- sion of-peripheral erythrocytes, and a ;nore zene/ethanol-treated mice was related to the severe marrow and splenic aplasia. In many dose of ethanol. Marrow granulocyte levels cases the severity of these cellular depres- from benzene/ethanol-treated mice did not sions was related to the dose of ethanol. The recover to the levels of air sham control val- effect of the ethanol seems to be one of in- ues which is in contrast to the marrow gran- creasing the hematotoxicity of benzene sins ulocyte levels of mice treated with benzene the majority of the observed blood dyscrasia alone. Splenic granulocytes were statistically are more clearly associated with benzene equivalent among all six groups regardless exposure than with chronic ethanol con- of treatment. sumption. ' ETHANOL INCREASES BENZENE TOXICITY 403 Orie reasonable explanation for the effects to be somewhat unique. Compounds known by the combined treatments is that to induce cytochrome P-450 such as phe- &mol, which contains seven calories per nobarbital and .3-methylcholanthrene in- gram, may replace dailycaloric needs with crease the rate ponnutritive[:'calores. However, based .on do not inc measured fluid consumptibn and average 1979). It has been reported that.there are food consumption, the percentage of weekly at least three different types of rat liver cy- caloric intake due to ethanol was calculated tochrome P-450 and tha " ', t. pretreatment with to be 4.3% for'mice-ingesting5% ethanol 4 ethanol, phenobarbital, or 3-methylcholan- days/week and 7% for mice ingesting 15% threne alters the relative amounts of the &iriol 4 days/week. These values are al- three forms (Comai and .Gaylor, -1973). most identical to those found in our previous Ethanol was reported to induce form I; phe- study (Snyder et al., 1981). The small ca- nobarbital, form 11; and 3-methylcholan- loric contributions of ethanol under these threne, form 111. It may be that`fo conditions are unlikely to produce a nutrient duces hematotoxins while f o r k s I1 I , produce more benign metabolites.tPerhaps for enhanced ben- form I produces lipid zene toxicity- i thanol increases the which would -`&d&nt levels of toxic benzene metabolites. Ethanol whereas forms I1 and could increase the activities of benzene me- uble hematotoxins which would -beA-chi- tabolizing enzymes within blood-forming nated before reaching appreciable kncen- organs producing, at these sites, hematotox- trations in the marrow. It may also be that ins in larger quantities or at faster rates than form I is more Drevalent L in the blo.oId-**fo..,r^&- would be produced by benzene alone. Chro- ing organs t-han _fo.rms I1 &d1.. ;II,-+I&2..SibI.n2.vLesti- nic treatment with ethanol has been shown gations of these and other hypot.hI esesawodd I. to increase the rate of benzene metabolism seem to be suitable for..further .work.s.u:-, I in vitro without greatly increasing the con- Finally, the appearance of tent of liver cytochrome P-450 (Sato et al., cells in peripheral blood is qu-&t 1980, 1981)."several investigators have pug- ferent from any of the pr&I ioS. ...;- * -..- gested that ethanol may increase metabolic tions of benzene.tox$.5ty .obse.r-v-e..A.d in thislab- '.-+*-?*,k-~ (m,activity by -mdifyihTg I .A. V.(. the .m, -1e11mb.,rane envi- oratory. The episode re ndopl&mic reticulum sents the second th Seeman, 1977),This mod- as many studies 0f.t ification may allow .benzene to migr >*`6,w>r- _ - I .-. .`... -. Cells mork susceptible to the toxic effects Of f 404 BAARSOl\I ET AL. CORNISH, H. H., AND ADELFUINJ,. (1966). Ethanol . ,potentiation of halogenated aliphatic solvent toxicity. mer. I d . Hyg. h s o c . 27, 57-61. CUR&, M.:AND:SEEMkN,-P:L( 1977). Alcohol t 0 h - . ance in a cholinergic nerve te&inal: Relation to the I .membrane ' expansion-fluidization theory -of ethanol action. Science 197, 9 10-91 1 DREW,R. T.,AND LASKIN,S. (1973). Environmental inhalation chambers. In Methods of Animal Exper- entation (W. I. Gay, ed. 01. 4, pp. 1-41. Aca- emic Press, New York ./ FRUHMAN, G.,AND GORDON, A. (1955). Quantitative effects of corticosterone on rat bone marrow. Proc S o c Exp. Biol. Med. 88, 130-131. GARRO,A. J., S~rrz,H. I. C., AND LIEBER,C.. S. ( 1981). Enhancement of dimethylnitrosaminemefab- olism and activation to a mutagen following chronic ethanol consumption. Cancer Res. 41( l), 120-1 24. GILL, D. P., KEMPEN,R. R., NASH, J. B.. AND ELLIS, S. (1979). Modificationsof benzene myelotoxicity and metabolism by phenobarbital, SKR-525A and 3- methylcholanthrene. Life Sci. 25, 1633-1640. GONASUN,L.M., W ~ E RC,., KOCSISJ, . J., m S w - DER, R (1973). Benzene metabolism in mouse liver microsomes. Toxicol. Appl. Pharmacol. 26,398-406. HIETA", E., KOWSAARI,V., LAITINEN, M., AND NORWNGA,. (1980). Hepatic drug metabolism dur- ing ethanol ingestion in riboflavin deficient rats. Tox- icology 16, 103-111. HILL, M. S. (1974). The effect of anesthetic-like mol- ecules on the phase transition in smectic mesophases of dipalmitoyl lecithin. I. The normal alcohol up to C equals 9 and three inhalation anaesthetics. Biochim Biophys. Acta 356, 117-124. KLMSSEN, C. D., AND PLAA,G. L. (1966). Relative effects of various chlorinated hydrocarbons on liver and kidney functions in mice. Toxicol. Appl. Phar- macol. 9, 139-151. KUTOB,S. D., AND PLAA, G. L. (1962). The effect of acute ethanol intoxication on chloroform-induced liver damage. J. Pharmacol. Exp. Ther. 135, 245- 251. LOBUE,J., DORNFEST, B. S., GORDON, A. S., HURST, J., AND QUASTLER, H. (1963). Marrow distribution in rat femurs determined by cell enumeration and Fe59 labeling. Proc. SOC.Exp. Biol. Med. 112, 1058-1062. LYON, R c., MCCOMB,J. A., SCHREURS, J., AND GOLDSTEIN, D. B. (1981). A relationship between alcohol intoxication and the disordering of brain membranes by a series of short-chain alcohols. J. Pharmacol. Exp. Ther. 218(3), 669-695. HIGH^^^MALING,H. M., STRIPPB., SIPS, I. G., .,:B., SAUL, w., AND WILLIAMS, M. A. (1975). hanced hepatotoxicity of carbon tetrachloride, th& .-acetamide, and dimethyl nitrosamine by pretreatmmt of rats with ethanol and some comparisons with ~ tentiation by isopropanol. Toxicol. Appl. P h a r m a , 33, 291. McCoy, G . D., H . S., KATAYAMA, s.,AND WYNDER, E. L. (198 1). Differential effect of chronic ethanol consumption on the carcinogenicity of N-& trosopyrrolidine and N'-nitrosonornicotine in malc, Syrian Golden hamsters. Cancer Res. 41,28dQ-,854. SATO, A, NAKAGIMA, T., AND KOYAMA, Y . (1980). Effects of chronic ethanol consumption on hepa& metabolism of aromatic and chlorinated hydrm. bons. Brit. J. Ind. Med. 37, 382-386. SATO, A., NAKAGIMA, T.,AND KOYAMA, Y. (1981). Dose-related effects of a single dose of ethanol on tbc metabolism in rat liver of some aromatic and chi* rinated hydrocarbons. Toxicol. Appl. Pharmacol. 64 8-15. SAVOLAINEN, H., VAINIO, H., HELOJOKI, M.. AM) ELOVAARAE,. (1978). Biochemical and toxicdogid effects of short-term, intermittent xylene inhalation exposure and combined ethanol intake. Arch. ToxicoL 41, 195-205. c.s.SEITZH,. K., GARRO, A. J., AND LIEBER, (1978). Effect of chronic ethanol ingestion on intestinal mc- tabolism and mutagenicity of benzo ( a )pyrene. Bio- chem Biophys. Res. Commun %5(3), 1061-1066. SNYDER, C. A., GOLDSTEIN, B. D., AND SELLAKUMAR A., et al. (1978). Hematotoxicity of inhaled benzene to Sprague Dawley rats and AKR mice at 300 ppm J. Toxicol. Environ. Health 4, 605-618. SNYDER, C . A., BAARSON, K. A., GOLDSTEIh. B. D., AND ALBERT, R. E. (1981). Ingestion of ethanol increases the hematotoxicity of inhaled benzene C57B1 mice. Bull. Environ Contam. Toxicol. 27, 175-180. SNYDER, R., UZUKIF,., GONASUN, L., BROMFIELED, AND WELLS, A. (1967). The metabolism of benzene in vitro. Toxicol. Appl. Pharmacol. 11, 346-360. STRUBELT, O., OBERMEIER, F., SIEGERS, C . p., AND VOLPEL, M. (1978). Increased carbon tetrachloride hepatotoxicity after low level ethanol consc:nPtion Toxicology 10, 261. WELCH, B. L. (1947). The generalization of stdcnu problem when several different population vanan- are involved. Biometrika 34, 28-35.