Tobacco Documents Online

THE EFF~CT~ Or CRA:.'GI~G BRANDS ON SMOIG~G BE~,~VIOUR EEP0~ U0. RD. 1409 RESTRZCTED | J 11.3.1977 £I~IORS: D.E. Cre~shCon P.H. Lewis ISSUED BY: R.E. Thornton PROG. EEF. 13.O1.04 DISTRIBUTION; Dz. S.3. Green Dr. Z.W. Hughes D'c. R.A. Sanfoz'd R..M. Gibb, Esq. R.S. Wade, Esq. E.G. lqlcholls, Esq. Herr R.E. Soclr.oz'f Dr. Y. Seehofer Mr. A.3. Kruxz'Fnskl I)~. C.J.P. de Siquei~a Dr. D.G, FeLt:on Libra~ Copy No. I, 2, 3, A, 5 el " " 7, 8 Itl It " " 10, 11, 12 " " 13, l& " " 15 " " 16 " " 17 " " 18 " " 19 " " 20, 21 mmemm. ',,.0 .-.,,j -.,,.3 BAT Co LTD - MINNESOTA TOBACCO LITIGATION

TABLE OF CONTENTS SUMMARY ZNTRODUCT ZON E:(~EEIMENTAL FRO C.,EDURE S RESULTS 1" EFFECT OF CHANGED DELIVERY ON SMOKING PATTERN DZSCUSS'rON I EFFECT OF CHANGED DELTVERY ON SI~KI:NG ]PATTER~ RESULTS TI EFFECT OF TTME ON SMOKING PATTERN W'ITH CHANGED DELIVERY C~GARETTES DISCUSSION II EFFECT OF TIME 014 SMOKING PATTERN WITH CHANGED DELIVERY CIGARETTES RESULTS Ill EFFECT OF MOTIVATION ON SMOKING PATTERN WITH CHARGED DELIVERY CIGARETTES DZSCUSSZON I~I EFFECT OF MOTIVATION ON SMOKING PATTERN WZTH CHANGED DELIVERY CIGARETTES CONCLU SION S RECOPL~4ENDATIONS FOR FUETHER WORK REFERENCES APPENDIX 1 3 5 11 16 19 22 24 26 29 30 33 34 BAT Co LTD - MINNESOTA TOBACCO LITIGATION O'-,

DCIPKLIIIAI 46D-6 Group Research & Development Centre, Br£t£sh-Amer£can Tobacco Co. Ltd., SOUTI~U~fPTON. llth March, 1977 THE EFFECTS OF CHA~GIZqG BRANDS ON SldOKDqG BEHAVZOUR (Report No. RD.1409 Restr£cted) SUMMARY A panel of eight male and eight female smokers were asked to smoke a med£umn£cotlne dalivery cigarette (1.4 ms) for about four weeks. During this time Can examples of the smoking patterns were recorded for each subject, using a smoklnK enalyser. Cigarettes were made freely available to the subjects throughout the period so that the medium delivery (control) brand was smoked exclusively. Half o£ the panel ware then changed Co a "high" n£coc4ne del£ver7 brand (1.8 mg) and half to a "1o~' nicotine delivery brand (Z.O sag). These c£Karettes were smoked exclusively for the nex~month dur£ng vh£ch time a further ten repllcatas o£ the smok£n8 patterns of each subject were recorded. The £iuaL phase o£ the exper~aaut :Lnvolved all subjects recucn£ng to the medium delivery brand and recording, over the next month, a further ten examples o£ the moking patterns. Analyses of these dace show that all smokers increased the volume o£ smoke drawn from the low delivery brand and Teduced the amount taken from the high del£very brand. All groups thus showed a marked degree of compensation for the changes in delivery. BAT Co LTD - MINNESOTA TOBACCO LITIGATION -...,,j

-2- The number of cigarettes smoked per day did not vary markedly when smokers changed brands, so compensation for changed deliver7 was essentially by the way £n which individual cigarettes ware smoked. Subjects used most of the mechanisms available to them to compensate for the changed delivery. A notable exception was that the butt lengths discarded when smoking the low delivery cigarette were longer than from the control cigarette; this was probably due to the longer tipping paper length on the low delivery cigarette. The panel was selected so that half of the smokers (4 female, 4 male) were o£ high "smoking motivation", as defined by e questionnaire ~ev£sed by Dr. M.A.H. Russell, and the other half of lower smoklnK motivation. Equal numbers of high and low motivated male and female smokers smoked each brand. Subsequent analysis failed to show any dlsnincnions of smoking pattern or degrees of compensation for changed delivery between the high and low motivated groups. Some differences can be seen, however, between male and female smokers both ~n the way in which the control cigarettes and those of changed delive~ were smoked. Analysis, of the smoking patterns measured for the first three cigarettes and lest three cigarettes smoked in each of the three phases of this experiment, shows that the changes made to compensate for both increased and reduced nicotine delivery were maintained for the period of the test. This suggests that the smokers' demands for nicotine hate not changed in this period. The design of this experLment (in which each brand was smoked exclusively for four weeks) will be used as the basis for future experiments, although £t may be possible to usa shorter periods of ~ma. The earlier approach to behsvioural studies (occasional smoking, only in the laboratory) has bean abandoned. BAT Co LTD - MINNESOTA TOBACCO LITIGATION O

-3- INTRODUCTION Zn most parts of the world the de•SEn of cigarettes is being modified and the dollveries of car (or TPM) and nicotine, as determined by machine emoklnK, have been declining. Although thfs trend has been particularly noticeable in recent years, with the introduction of "league*' cable• of cigarettes ranked by deliveries, the reduction in deliveries has been going on for some time; e.8. Todd (1)p in a survey of U.K. smoking patterns records that, compared to some cilarectes of 1935 vintage, the 1973 average was nicotine delivery 47Z and tar delivery 57Z, Such trends are particularly marked in developed countrles; e.K- in the U.S.A. c~Karettes with tar deliveries of lOmg or less have been introduced (Merle (9 m&), Now (2 rag), Carlton (5 m&), Carlton 70 (I rag)) and in some countries upper llmiCs for car have been voluntarily agreed by manufacturers. IC is often assumed by smoker• and implied by ochers that there are advantages in changing to a brand which offers lover dellver~es of Car and nicotine and to an increaslnK extent such advice is hero•inK more expliclt e.g. "Smoke a brand of cigarettes in a lover 'tar' Stoup than the brand you smoke at present and aim progressively to reduce •till further" (2). However there ks • considerable body of evidence that: smokers alter their imoklng behavlour when smoklnE different types of cigarette. Evidence has been presented (3) that smokers may •how • degree of compensation for changed nicotine delivery 8o that they attempt, within llmics,to equalise the nicotine delivery, irrespective of the effect this might have on the tar and gas phase deliveries. As pointed out by Russell (4) this means that: compensation for a reduced nicotine del£very can cause an increase Ln the ~ntake of Car and carbon monoxide. BAT Co LTD - MINNESOTA TOBACCO LITIGATION O

-h- 6 • Much of the evidence for compens•tion, quoted ~n (3), has been obc&ined under labor•tory conditions and has involved ~omp•rlsons of s£ngle cigarette saok£ng measurements. It is, however, poss£ble thac a smoker can change his behaviour with time and become accustomed to a brand offering higher or lower deliveries, espec£ally if he smokes the 'new' brand exclusively. Thus, ic has been suErs•ted in • U.K. advertising camp•fEn, by the manufacturers (Call•her) of a low dallvery cisarac~e (Silk Cuc) that, after 8moltoK ch£s ciKue~ce for three weeks, the smoker would be convinced of the merits of the product. In view of the considerations ~isted above, it was decided Co carry ouC • behavloural experiment of considerably greater scope than any previous study made at Southampton. Smokers of a medium delivery cigarette (~ I.A mg nicotine in the U.K. mark•C) would be monitored for • period of about 1 month and then asked Co smoke, for a aim/lit period, • brand with a substanCe.aLly different nicotine delivery (~ 0.8 mE for half of the panel, ~ 1.8 uK for the ocher half of the panel). For the . final period the smokers would revert to smoklnK the brand dally•tinK 1.4 ~ nicotine. It is to be expected that there ~s a Kradatlon of demand for nicotine among smokers. Russell ($) has estimated chat some 20Z of ~nokers are not particularly interested in nicotine an all. Russell has ass••bred a e questlounaire (which he states to be a combination of questions devised by ocher workers) from which he has sugsesced • relationship between factors isolated from the replies to the questionnaire and nicoclne demand. IC therefore seems plaus£ble thac h£Khly motivated smokers (as defined by Russell's .quesCionnslre) would be the most likely co show a BAT Co LTD - MINNESOTA TOBACCO LITIGATION w ",.C)

-5- hiKh degree of co•pens•riot for reduced delivery. The more highly motiv•ted smokecs u~Lsht •leo be expected to show the more st•ble maintenance 0£ their h•bit with time when they thanked to • lower delivez7 brand and possibly to show • more precise re,teflon of nicotine intake th&n the less h£Ehly motivated subjects. The panel was therefore selected zo contain equal numbers of hi@h and low motivated male and female smokers. The objectives of the study ware: (a) To study the effects o£ oh•need delivery on smoklns pattern. (b) To study the effects o£ time and thanked dellvery on smoking pattern. (c) To study the effects of motivation (ss defined by Russell) and changed delivery on smoking patten. EXPERIMENTAL PRECEDUEES The Panel The smokers selected for this test were e4Kht male and eiKht female smokers, •11 of whom had partlc£p•ted in previous experiments in ~h£ch smokLn@ patterns were recorded. All subjects were thus fmnil£ar with the •ppacatus, procedures and env£romneuc used by us for such measurements. After the survey of 81oklng patterns with£n Group 1~. & D. Centre (6), all smokers completed the Russell Questionnaire concerned v~th smokLn4 motivation (7). The scores for £•ctors ~, (scLmul•t£on smoklng), V (eddictive 8mokins) and VZ (automatic smokinK) (Russell's teminoloKy) were used to derive • smoking motivation score. For practical re•sons it was not possible to select •11 the subjects "with the hiKhast and lowest motivations, so an element o£ compromise w~s necessary. Two groups BAT Co LTD - MINNESOTA TOBACCO LITIGATION O (..r'i ---,d Oo

-6- of four males an~ four fmlas w£chin the panel v ere dlstinsuished as having higher and lover than averase motivations, as idencifled by l~ussell's questionnaire. All the subjects selected for thls cast normally chose Co smoke Embassy reKular lenKch filter tipped ciKaretCea or a cisarette wlth similar delive~ characteristics. Methods Smok£ns patterns vere recorded by a smok~nK analyser (8). For test purposes all cLKareCCes were smoked in a controlled enviro~enC room (9), whilst the subjects listened Co Cape recorded music of Cheir choice. The smoklnK pattern dace, recorded on maKneClc tape, were translated Co reduced format paper Capes and read on a teletype machine. From these printed rec6rds the primary results were calculated. As the dace are scored on masnecic cape subsequent dupl£cacion of ~he m~kinS patterns is possible. Cigarettes The ciKarectes choseu for the trial were: Embassy r~tnSs F~nbassyKinS Size F~nbassy ExmraKild U.K. Home Trade B.A.T Souchmnpcou U.K. Home Trade The physical dimeusions sad analyc£cal ~atawhen smoked under 8tandard 8 conditions are shorn in Table 1. Th~oushouC this zapozc the cenns concroX or med~umde~ivaz-y c£sawacce rater to Embassy KLnKs, hish delivery to EmbLssy Y~nS Size and low dalivary to Embassy Extra ~ld. BAT Co LTD - MINNESOTA TOBACCO LITIGATION ',,O OO

-7- TABLE 1 FlffSZr.,AL D.Zt~NSZONS AND A~ALYrZCAL DATA FOR T1~ CZGAX~TES ~l:~H St'tOteD UNDER ST/~IDAI~D CONDZTZONS Overs11ZenKth Rod length circua~ere=ce Filter ~ter£al mouth tobacco F£~ter length mouth tobacco Overv=ap Ventilated PO range (17.5 aI/sec) TFM Total Nicotine &lkaloids C-,,,) (.==~ (m) (--,,) C,==) (:m) (o~ W.G.) (uS) Embassy i~ngs 83 65 24. $ Cellulose Acetate My~ia 6 12 21 Bo %2-14 23.0 1.4 Embassy King Size 82 68 24.5 Cellulose AceCate m 14 23 No 12-1& 36.0 1.8 Embassy Extra Mild 83 64 25 Cellulose Acetal:e Hyria 6 13 24 Yes 12-14 16.0 1.0 These cigarettes ware selected as being typical of the range of ciEareCtes avaIZabZe on the U.K. market at pres~c. The brand Embassy King Size, ~x 3.A.T Southampton, was included sLuce this made It possible for all three brands to*have a deEree of S/=L~larity, i.e. all three were Eabassy variants. All ciaarettes mnokad by the subjects in the laboratory were selected to be within the d:m~ :esista~ce ranKe 12-14 cmW.O., when~asured ac a £Zav race of 17.5 ml/sec. The cigarettes were not weight selected. The BAT Co LTD - MINNESOTA TOBACCO LITIGATION r,.,,,a m

-8- clsareCces issued co the subjects co smoke outside the smoking behaviour laboratory ware in sealed packs of 20. Experimental Design Phase I" All subjects were given the medium delivery cigarette co smoke exclusively for four weeks. During this c/me each subject was monitored on can occasions smoking an example of chls brand. The spacing of recordings was as even as possible ChrouEhouc the period; half of the recordings were oaken before lunch and half after lunch to mlniJnise any. effects due to C~ne of day. Phase I~ Half of the higher motivated subjecus and half of the lower motivated subjects chahged to ~he high delivery brand and the remaining subjects chanSed to the low delivery brand. & further tan smoking pattern measurements were recorded ~rom all the panel members evenly spaced tbxouKhout the next four weeks. The changeover was arranged so that the flrsC example of nhe replacement brand was smoked in the laboratory and recorded. Cigarettes of the replacement brand were also issued Co the subjects ad libicum. Phase ITI el|l £11 the smokers returned to Chemsdium delivery brand. A further ~enmeasuremant8 were taken over the four weeks with. as far as possible, equal spacing between records. The first c£KaretCe smoked by each subject of Ch£s f/hal phase was recorded and, as before, subjects smoked the brand exclusively. BAT Co LTD - MINNESOTA TOBACCO LITIGATION w %,0 CO .~b'"

-9- A record yes kept of all clgaretces issued throughout the three phases of this experlm~nC so ChaC some £ndicaClon of changes in cigarette consumption was available. When a subject had finished smoking a cigarette in the laboratory the cigarette holder containing the stub was placed in a tray cont&in4ns granulated ao1£d carbon dioxide. The length of the buCc was unassured and recorded. The ££1ter was then cuc from the c£gerette butt and analysed for nicot£ne by a GLC method (10). DATA HANDLING The smok£ng mnalyser records I00 measurements per second during a puff; 50 of which are pressure readings and 50 of which are flow readings. The muounC of data recorded from the 480 records of smoking pattern occupies over ZOO0 metres of magnetic tape with a packing density of 800 BPI. IC is noC pracC£cal Co axan~ne every facet of such an enormous amount of data, so the data ware reduced and condensed co 1ore m~naseable proportlons. This report ks based on an analysis of the data obtained from reduced format paper tapes, which have been transposed co separate sheers of tables and analysed manually. Even this reduced dace sac involves the mmnlpulaCion of some 2&,OOO basic dace points (ZnCmrval. Volume, Duration, Total Pressure to Draw Puff, Puff Number, Butt Length and Hicoc£ne retained in filter tip). After manual calculation of the data had been completed the reduced dace sac yes copied from the cables to punched cards. Programs were developed ~o combine the data in varLous ways Co confirm the results of the manual .analysls. The dace ware BAT Co LTD - MINNESOTA TOBACCO LITIGATION ",,,.o Oo

-10- then amalysed by an analysis 0£ variance teclm£que to show the statistical s£Kniflcanee o£ the chanses. Further deta£1s of the statist£ca~ analys£s are shown in the Appendix. The~e weme three main objectives o£ the experiment wh£ch have been dealt vi~h separately, as Sections Z, 1Z and ZIZ. The data used in each section £s the slme but has been combined in dL£ferene ways to answer the speclfic questions posed by the objectives. BAT Co LTD - MINNESOTA TOBACCO LITIGATION I m 02)

-Ii- PART ~ - EFFECT OF CIIANGED DELIVER~ ON SHOICXNG PAXTZ]~ RESULTS I Table 2 sho~8 the average results over all subjects and all replicates for each of the three phases of the e~perimenc. Table 3 sho~s the statistical significance of the changes be~een =he phases. A new method o£ m~slysLs Co show the degree o£ compqmsetion £or changed del4ve:7 is also presented, cogeche= with the results obcaLued (Table A) using Chls method. BAT Co LTD - MINNESOTA TOBACCO LITIGATION I k..rl

-12- TABLE 2 EFFECT OF CHANGED DELIVERY ON SMOKING PATTERN Pauramecer Tonal Puff Vol,--a (ml) AveraKe Puff Volume Tor~al Puff DuraClon ~Averase Puff DuraClon (rod (IBC) (Sen) Tonal Interval benwten Puffs (sac) AveraKe Interval between Puffs (sen) Total Pressure Co Drew Puffs AveraKe Pressure Eo DzawPuff8 AveraKe Number of Puffs ~verase Bunt LenKch Estimated Number of CiKaretce8 Smoked per Day Nicotine Delivery EsCisuaCe~ frm Tilter Tip Analysis (era W.G. nc) -(cm ~.C. nc) (--) (~) De I ivery in Phase IT l~gher Lover K£Kher Lower Higher Lover Higher Lower RiKher Lower HIKher Lower H£sher Lower R~Kher Lover Phase I 545.8 570.5 43.8 45.2 28.6 30.3 2.3 2.4 439.1 437.0 38.5 36.5 526.3 554.9 42.4 44.8 Phase II 433.4 624.9 39.6 53.4 24.0 32.0 2.1 2.8 371.4 418.8 36.5 37.2 449.2 527.3 41.5 45.1 HiSher Lover B/Kher Lower HLshec Lover Hisher Lower 13.0 12.6 31.4 31.2 33.2 31.1 1.7 1.8 11.5 12.0 33.g 32.8 29.9 32.5 2.0 1.4 Phase XII ,, | 534.4 522.4 44.5 46.1 28.7 27.1 2.3 2.4 423.4 431.1 39.2 40.9 519.6 516.2 43.5 45.4 12.6 11.4 30.7 31.5 33.3 32.8 1.5 1.4 BAT Co LTD - MINNESOTA TOBACCO LITIGATION m ',,,E) OD

-13- CLo~eL of Si~ific~ce 0.10) Ps~m~t~ l~[f Ikm~er. I~tt Igdmsch L£t Drme beLsumce Totm2 TAm A~iSht Tip N£¢ocine EJt JLmac~l uicot~e DelLvery To~al of Pressures to ~sv Puffs &verqt rreaJuce Tot~ Volume of ~uffs A~reZe VoLm~ TotaZ ~rscLon of 7~fe AverSe ~ati~ TotsL ~te~aI &Terse XKcereal ~UP 2 ~se ~ ik~ she 2~ delAv~? *(~8c~ Is ~se 2 in delive~ TtLJt£vt ~ re~8~Lve ~ uHm to ©out~ol ~LEarel:te ~mSer ~r Lom~ l~ 2 r3um l ~e 2 to )~ ao ~um~e lle ~hanEe R~r ~ser ~,n~ger Ik, chaa~e /~ase X ~o 9: no ~m~e No ~um~ S~ c~m~ No e~mSe No elumXe No ~mse ~r No ~m|e ~mr m~ . . BAT Co LTD - MINNESOTA TOBACCO LITIGATION (,..j'l %O

The deEzee of compensation for chanKed delivery of nicotine may be calculated in the £oIlowlns way: let Ox ÷ OZZZ be the average measured intake of nicotine in 2 phases I and IXZ. lec EXZ be the expected in~ake of nicotine in phase IX if that e£KaretCe had been smoked £n c~e same way as £n phases Z and IlI (i.e. no compensation for changed delivery). let OZZ be the measured intake of nlcotine in phase XZ. Then the percentaKe compensation is defined as: (Oii -EII) IOl ;011~- EZT/ X 100 - Z compensation. where the value of EII is calculated using: z OI + OllI" where SII end SI +~II are the deliveries obtained by machine smoking 2 under standard conditions of the particular brands beinE examined. From the formula to calculate delree of coupensatiou it can be seen that i£ the observed intake (Ozl) of nicotine in phase 11 ~s equal to the utinutted (Ezz) /:atake of nicotine the percentese compensation is zero. Z£ the observed £ncake (Oil) i8 equal to the averale measured intake dr (i.e. perfect equalisatiou). This method has been applied to the daily nicotine intake (delivery peer eisarette x number of e£Karetcee smoked per da?) and is shown as Table &. BAT Co LTD - MINNESOTA TOBACCO LITIGATION O (,..,y'! ',.,O

-15- TABLE 4 EFFECT OF CHANGXD DELIVERY ON DEGREE OF COI~ENSATION FOR DAZLY N~COTZNE INTAKE EXPRESSED AS A PERCENTAGE Subjects Male Hale A11 All nel£very in Phase XX R£gher LoweY OI ÷ OilI EZI OXZ 2 (mE) ('~S) ii 59.6 77.8 60.3 50.1 45.3 54.7 53.1 57.X 43.9 55.3 65.4 66.5 33.O 43.7 Perceutase Compensanion ÷113.4 ÷ 6g.5 - 7.0 ÷ 87.0 BAT Co LTD - MINNESOTA TOBACCO LITIGATION

-16- DISCUSSION OF R~S~LTS - It can be seen from Tables 2 and 3 chat, in general, smokers have sho~n a degree of compensation for changed delivery. The attm~pts to compensate have £nvolvad almost every mechanism that can be used on a G single cigarette. The number of c£garetces smoked per day was not: changed markedly. It is also worth noting that the human smokers have taken on average about twice the volume of smoke from the control clsarette as vould be taken by a machine operating under standard eondltlons. The interval between puffs was about half that taken by a machine. Pu£f durations and the number of puffs drawn v~re also greater for the htuuAn subjects than the standard s:oking conditions. l~eference to Table 3, the scaCletical evaluation of these results, shovs that there vere a fev differences in the way 4n which the male and female subjects smoked the cigarettes. Generally ~en took a miler volume of smoke per puff but more puffs per c~a=aCte than men. The pressure used to drav the puffs was lover; hence the sverage lit draw resLstance exparienced by female smokers was less than Chat experienced by male smokers. The degree of compensation for daily nleot£ne ~ntake (Table &) shove that the male subjects were closer to equallsaclon vhen smoking the high delivery cigarette Chart the low delivery cigarette. The female smokers shoved better regulation of daily nicotine intake for the lov delivery cigarette than the male subjects but virtually no compensation for the high dellvery cigarette. The levels of nicotine drawn from the cigarettes, calculated grom the amount remaln£ng in the filter tip assuming filtration ef££ciency BAT Co LTD - MINNESOTA TOBACCO LITIGATION N m

-17- to be constant, are within the range of deliveries suggested by the results of Project WHEAT (Ii). Zt should be remembered, however, chat the results of the present study are object£ve measurements made from cigarettes smoked by human subjects in a laboratory. The conclusions drawn from Pro~ect WHEAT are target levels ~Lch should be achieved when the cigarettes are smoked by a machine under standard conditions. ~, however, human subjects smoked in a s~ilar manner to the standard machine smoking conditions, them deliveries would be s4,n£1ar. Zt was shown previously (6) that the volume og smoke taken by human smokJrs £rom cigarettes, with a n~¢ot~ne delivmry og 1.68 uS when smoked by machine, was AO6 ml. This demonstrates that £or clgaret~es with, by present standards, hlgh nicotine deliveries, human and machine smokir~ need not differ markedly. Although smokers have increased the intensity o£ smoking the lower delivery cigarettes and decreased the intensity of smoking the higher delivery cigarettes, they do not appear to have equelised deIivery vlth the control ciKarltte. Rowever, the n~cot£ne daliver£es quoted for these ¢isarettes should only be taken as approximate because It is known that the velocity of msoke passlng through a filter ~£ects its filtrat£ou efficiency (12), and the relationship is not linear. If the total volume of smoke drawn from a clgarette is divided by the coral duration of the pu£fs an average f£gure for the flow rate can be calculated. For a cigarette with a diumeter of O.8 um the velocity of the smoke is (numerically) approximately double the flow rate. Using the averale veloClt~.es meutsured in l:his experiment, the actual £illmration e££ic£encies of Che £ilcer ~ps analysed may be within ±IOZ o£ thac calculated ~rom BAT Co LTD - MINNESOTA TOBACCO LITIGATION

-18- standard machine smoklnE conditions. As it is intended to duplicate representative smokir~ patterns of all the mubjeets used in this teat. ic is hoped Chac a better esCixuce of actual deliveries to the smoker will be obtained. One of the mechanisms by which a smoker mlsht be expected to compensate for reduced delivery is by smoking a greater number of cigarettes per day. S4milarly it might be expected that fewer of a higher delivery brand would be smoked. The evidence collected during this study suEEests that the number of cigarettes smoked per day is reasonably constant, i~respectlve o£ the delivery of the clgarecte. Zt should be made' clear, however, that the cigarette consumption figures quoted are approximate. Cigarettes were issued to the subjects on demand. The figures quoted will include those given away. It is also possible that smokers who ware offered, in affect, unlimited supplies of free c4garectes might have abused the system and become more generous or even stockpiled cigarettes. IC is to be expected that such abuses would not be large as all the participants in this experiment normally receive a weekly issue of cigarettes. BAT Co LTD - MINNESOTA TOBACCO LITIGATION m

-19- PART II - THE EI~"ECT OF TIME ON S~.IOKIIqG PATTERN WITH RESULTS - II A smmnary of the average results Ls shown as Table 5. This table has been constructed by the 8umunation of the results obtained for the first three cisarattas smoked in each phase for comparison win the results of the %ast three ¢iKarettsa mBoked in each phase. Addltionally, comparlaon of the last three 8moklnKs in Phase X ~r£th the first three in Phase ZZ shows the chanses in smokin$ pattern which are associated with the thanes of ciKarecte. Similarly, a comparison of the last three cisarettes smoked in Phase ZT ~th the first three in Phase T~I~Z shows the changes in smokinK pattern when returning to the control brand. Due to the desiEn of the experiment, in which smokers were monitored in the laboratory on a~ternate working days, the comparison is not 8tEietly between week 1 and week & of smoking a particular brand. This is because subjects were measured smokinK 3 ciKarettes in the ~aboratory for one week then two the next, Because of this desiKn the thlrd replicates for half the panel were in fact ueasured £n week two, and the first of the flnal three replicates was measured in week 3 for half the panel. A statlsti=al evaluatlon of these results is shown as Tables 6 and 7. BAT Co LTD - MINNESOTA TOBACCO LITIGATION ",,.o

-20- ~A,~z_____~s K3r~¢l" 01r ~ OH SI4Q~IIIG CXG~UL'I"T~ ~ ¢;IAIIGZD lI13"gZILY P~mter To~sl pt, f[ ToLume Av, ttqa M! Io1~mt TocaX l~g Dutaclon Avtrqa M! hcitiem .~ar.al X,,tmaLs betveeu Puffs &varaae Xs~e~waX be~,eem Pugfa Tar.81 o£ Pressures to Drev IhaZZa AI, traae of l~essuTes eo Drav Puffs AvaraKe Number at ,uffe &vmcoSe Ikstt Letti~ch AveraSa Eaeimated liLeotin8 DelivetT (ml) (11) (e-c) (s.c) (ue) (jet) (era IhG, ~) (cs g.G. oee) (ms) (uS) n D411vez7 Yirst Thr~e Claaretces Ln ,base XX ,base X Phase X1 i~haje ~X Riffler 5:32.9 452.6 572.2 Lover 554.7 650.6 566.$ ELsh.~ ~. a 60.& &a. 8 66.!) 54.5 68.6 Iltsh~" 27.9 Z4.5; 25;. 9 1lamer 30.7 3:3.6 25).0 ltiahlr 2.1 2.2 2.3 Loe~r 2.$ 2.8 2.6 1Liaber 6&5.2 :385.1 ~0.9 X,m~= ~1.S &14.2 429.0 ]lAl~az 36. :3 37. & 37.6 35.8 3S.8 &O. 9 EAsber 530.0 412.O .5a4.4 l~m,t= 5.5:3.8 .545J. 3 .~29.4 WAsher 60.$ 62.6 42.8 la~er a&.8 4s.g 46.8 Iligher 1 $. S U • 1 113. :3 lame:: 12.5 il.3 11.4 RISbem 31.7 34,6 31.O I,mmr 31.3 33.8 32.3 ItiSbeT 1 • 6 2.0 1.1 Lovur 1.8 1.5 1.8 Last Tln'ee CLf.azettes Phase lr Phase XX Phase 1111 I .576.2 420. L 506.6 35&.g 627.2 St3.& 47.2 39.0 6:3.8 &?.2 53.6 6S.2 15;.6 7.,3.0 27.7 Z9.4 31.8 25.8 2.6 2.1. 2.3 2.a 2.8 2.3 425.1 3 S1.7 428.6 4:36.5 405.5 420.8 38.2 .96.1 61.& 36.2 $&.0 tO.$ ~39.6 432.8 493.2 .56.5.9 .530.0 697.6 &4.6 ~.S 63.3 65.6 65.5 &3.? 12.51 11.3 12.2 12.6 1:~.1 11.$ 31.4 33.4 30.3 31.1 32.6 30.9 1.8 Z.0 1.3 1.S L.6 1.2 BAT Co LTD - MINNESOTA TOBACCO LITIGATION ",,.O q

-21- TABLE 6 SIGNIFICANT D!FFERF~CES BETWEEN SUB-PHASES FOE G~OUP 1 L (CHANGE TO tlXGl~ DELIVERY) Parameter 1-3~ 4-3 Number of Puffs Butt Length Lit Draw Resistance Total Time Alight Tip Nicotine * Estimated Nicotine Total of Pressures to Draw Puffs Average P=essure * Total Volume Average Volume ** Total Puff Duration Average Puff Duration ** Total Interval Average Znterval e. e, 4-6 6-7 t* e. e. 7-9 ~rA i-9 ° %__. TABLE 7 Szc~zFI .c~ DX, FFE~ENCES BETWEEN SUB-P~SES FOR C~OUP 2 ~CHANGE TO LOW DELIVEKY) | ParmnCer |m ~umber of Puffs Butt Length Lit Draw Resistance Total T~neAlight Tip Nicotine Estimated Nicot~a Total of Pressures to Draw Puffs Average Pressure • Total Volume Average Volume Total Puff Duration Average Puff Duration Total Znterval Average Interval 1-3 ~-3 ee Qe ee e. 6-6 6-7 d, ee e 7-9 1-9 dnV * Significant at 0.10 level. ** Significant at 0.05 level. i % See appendix for details of sub-phases (page &O). BAT Co LTD - MINNESOTA TOBACCO LITIGATION

-22- DISCUSSZON- ZT It can be seen from Tables 6 and 7 that there were some stat£stically s£Knlf£c•nt dlffezences baleen the first three and the lest three ciKarettes smoked in Phase Z. Similarly there were a few statistically 8iKniffcant differences between the first and Final weeks of Phase TTI when the subjects returned to the control delivery cilarette. There were no stat£stlcally siKn~f£cant differences, however, between the £1rsc and last three cigarettes smoked in Phase IT. This flt,dlng applies to both groups of smokers, who thanked to either the higher delivery cigarette or the lover delivery cigarette. There were statistically significant changes observed for uoet parameters between the last week 8moklng the control cigarette and the first week meok£nK the changed delivery ci8•rette and between the last week smoking the changed delivery cigarette and the first week o£ the return to the control brand. These findings •re interpreted on the basis that the smokers changed most aspects of their smoklnK patterns in response to changed delivery cigarettes and maintained the modified smokinK pattern for • period of four weeks. This £indinI is •t variance tr~th the popular belief that mmkers modify their smok~n$ pattern over • period of weeks in respoeme to changed delivery. The present results suggest t~st •mokers cmnpm=sate got changed delivery as soon as they start to smoke • changed delivery braml and malnta~n the new smoking regime. It would require another experiment to determine the length of rime for which the modified smoking pattern perslsted. The supposition that some smokers can become used to • lover nicotine intake in time assumes that" their demand for nicotine ¢8n be reduced. This ImplLes that the metabol£ma of nicotine can also BAT Co LTD - MINNESOTA TOBACCO LITIGATION ",,O

-23- be reduced at about the same rate. When changing to • brand offering different delivery, it is also ntctssary for ~he smoker to modify the physical acts of smokinK ~d~Lch Involve the r£tuals of c4Karette selection, puffins and ~nhaliuK. These responses may well have become Pavlovian in established smokers and difficult to modify in • ¢omparetLvely short ti~. BAT Co LTD - MINNESOTA TOBACCO LITIGATION m m

-24- PAET Z'rl" - TF~ EFFECTS OF SMOKING MOTIVATZON ON SMOKL~G PATTERN WITH CHANGED DELIVERY CTGAEETTES RESULTS - I~I A eumBary of the average resulns is shown as Table 8. The male and female smokers have been combined to occupy four groups; to accommodate the two levels of mot£vat£on and the two levels of chanEed delivery. A mtatlstical evaluation-of the ~c£vation effect is shc~n in colin,- 3 of Tables 9 and 10. BAT Co LTD - MINNE..,C~TA TOBACCO LITIGATION m m

TASL.t 8 IDTECTS OY NOTZYATT.ON ON SHOKLNG PAT~Im UZTH CIIAJG~) DELI~E1tY ¢ZGAP2,'TTES DeILve:7 l~mrmsmtetr l~r.£vaticm In Pbaae Z Phaao ZZ "focal Puff Volume (ml) Irish Iti|t~ 612.3 I~sh Lov 502. 1,m, |LSh &?8.9 . Lmw Lm, 638.6 &ve~rsKe Pu~f Voluae (ml) KLSh ILLsh 62.9 u48h r, ov 63.3 . Lmv ItLsh 11.3 Lau /dyv &7.3 Tots1 Du~stLou oE Puffs (eec) 18LSh Rip 32.6 SLSb Lm, 2S.t 1Wn, ILLSb 24. S Low ~ 31.6 Avzralre kwacLon of* Pui',fl (sac) lush IILsb 2.28 KLSh Lay 1.*~ Lov Nip 2.11 taw raw 2.3& Tots1 oE Znr.ez~&le be~voen 1P~££8 (0o¢) ILLwh uclh 390.2 ILLEh Lay 513 • 6 Lee ELp 685. ! Lag Lou 360. $ Averxlr~ IJtterval betweem 1~t£8 (e~u~) tULsh lUSh ZI.3 Ilisb Low 68.6 Low mSb &6.1 Lay Law" 28.8 1~er~Z e£ FresmJ/es I:o Dirty l~dEf[s (ore U.G. oe~) 1~81b lush 54J.1 B,LKh Lee 473.9 :m, ulsa 66~.4 Lm, 6.16.O Aveca4De Preseu=.s to Drw l~d~£s (at W.O. oec) ILLsh ItLsh 41,L ~ ~o.~ Lov ~Sb 40.0 Law ~ 47,I &veraSe Nmnber of" Ih~t8 ILLsh 9U4b 14.:3 nish Lee lZ.S /am Rish 11 .& Z.me I.$.5 &versKo ki:t Lensth~) ELsh ILLsh 32.6 u48h Lay 28.5 Lmw uLSh 30.5 " Lov l~w 33.J ave:faSt Ir.Jcimaeed lli~,ot£na htlvesry (niL) BiSb Bl8h 1.85 nlsh r.~ 1.8y ILLsh 1.81 Lev Low L.5)$ 4nreral~ ~stf~a~ed Ilumber of I[ISh ll~l~h ~1.~ Cilr~req~eo Smoked per D~y IEil~h Loe 30.0 1KLSh 33.9 ~ 31.6 Phase ][I Pl~xe Z£Z ~K)0.O 611.8 529. r. 423.6 $66.7 456.9 720.2 621.2 38.5 43. I 68,1 4,2,B 36.7 4L.~ 55.0 a~8,Z 28.1 34.o 3z.2 25.0 1|.5) 23,$ 32.8 25).Z 2.16 Z,'ts) 2.84 1.$3 1 • 9g 2. L4 2,S0 2.28 32O.7 3NbO. 8 513,1 535). o 422.0 686. t 324,6 323.3 28.7 27.3 51,3 60.6 ~.Y 48.6 26.| 27.2 SOB,~ 59L.7 636,7 6OO.6 3~0.1 6~7.4 817.5) ib31.8 39.1 41.7 39.? 40.5 39.0 60.7 47.2 69.0 13,0 14.2 11.0 5).9 10.o 11.0 13.1 12.9 3.5.2 31.6 29.2 ~7.3 32.7 29.7 36.3 3.5.8 2,27 1 .S8 1.35 1.28 1.81 1.36 1..52 1.69 28.9 31.5) 32.6 13.7 30.0 3&.S ~2.0 31.4 BAT Co LTD - MINNESOTA TOBACCO LITIGATION m m

-26- DISCUSSION -III The l~ssell Questionnaire (7) was scored and the scoEes for factors I (St£u~lation), V ~dd£ct~e) and VZ (Automstlc) combined. These factors are considered by 1~ussell to =elate to the meed £o: n£cotine in smokers. It may be interpreted that the subjects wich high scores are more mocivated to smoke for the nicotine delivery of a cigarette :ha= subjects with lover scores. The maxJJum score £s 39 and che avereKe scores for the panel vere: HiKh Motivation Group. Low HDtlvatlon Group Hale F-~-le 16.75 20.75 8.25 10.25 bssell's P opulee£on (AveraKe of 175 subjects) 7.g The panel yes thus composed o! Jokers 1rich average and higher ~%8n everase motivac£onal scores. The female members of the pamel had sllghtly hisher mean scores than the male members. Group I. & D. Centre has a preponderance of hlghly nm~Ivatsd smokers accordinK to Russell, even tho~sh those vith the highest motLvation could not be used in the present ezperlment. BAT Co LTD - MINNESOTA TOBACCO LITIGATION w r,,o

-27- D SZGIJZ/'/C~L4T EYlT.CTS ~ atOU~ 1 Ol'rO! ~I.2VEI~' ~ PII~I 2) Deliver7 SeT I U;~VatL~ a i Hi LtLL~ Q ttl ttint lill Delilit7/ lliivacioa Set/ li£vsi~ llivsi~u t |ai~oil Piopie ltiil Rill tit tli ill lersoa/ ~live~ lllt Ill ttt ttl oil it/ llt s=izr~ ZTIi~s ~ ~ z t~e u~Ivui ix ~ !) PltINeer ~iiviry i if ~lis kAi ~th I~P-! Ill lille " v il Alibi Tip Ileol/ll ~ll lico~lo Toil # ~llll IO Dirt its Ill ~ill ilill tl ieTqt ~ii tlEi llil biervai dm~lle biafra1 Q O dltl W m llt O i t * 8~tLumt si O.lO lovih llt I~ticini It O.Ol leveh t lla • . ~ltvary/ iflvi~gon Delivery/ lien/ : ItLvilioi II Dell.arT! l/ mlt£~&t~ ttivalim~ m a o o e dr lloplo lit tit all lit tit ill lit 15tl People/ ~Livt~7 t I t INbl d~lbl Oti ata a~a dlt.a ,l BAT Co LTD - MINNESOTA TOBACCO LITIGATION OO

-28- The results of the statlstical analysis (Tables 9 and 10) show that there were very few significant dlffexences between the motivational groups. There ware no differences found between the hlgh and low motivation groups who changed to the high dellvery cigarette. Differences, vh£ch were just slgu£f~caut at the 90% confidence level, were found for the number of puffs, Coral pressure used to draw the puffs and the average interval between puffs for the moC£vaC~onal Kroups who changed to the low delivery c£garetta. Reference to Table 8 shows chac the subjects vlth low mot4vat£on took the greater number of puffs, shorter ~nCm~vals between puffs and higher pressures to draw the puffs. This effect can be seen in all three phases of the exper~nent. Th~s is a reversal of the slcuatiou that yes predicted, as it was expected thac subjects with a higher motivation Co smoke for the nicotine content of smoke would show a greater intensity when smoking low delivery cigarettes. IC would also be expected chac compensation for reduced delivery of nlcoC£ne would be more complete, i.e. nearer to aqual£sacion ef dellvary, for the subjects ~r~-Ch h£gh~ moc£vatlon. The results of thls test, however, ahoy that the differences ~n thm n£cocinm dal£vmry of the c4garatces have had a marked affect on the degree of compensation whereas the different mocivaclonal groups as ident££iad by ~ssell's quesc4onnaire, do noc appear to show any lutmrprecable differences. The small size of the motivational groups used in this axper£menc and the large variations between people within these groups may be distorting cha observed effects due co motivation and c.ot~cuZis~ ochers. Zn general experiments duisned Co study the differences between ~coups of people require a large simple of peopZa from each group. BAT Co LTD - MINNESOTA TOBACCO LITIGATION O k.r'i w OO

-29- COI4CLUSI014S These results show that, KsneraLly speski~, the smokers in this panel coupenaaCed for chansed del£vet£es usins most and sometimes all of the methods available to them simultaneously. The number o£ cigarettes smoked per day did not alter markedly, ~hile the average butt Lengths discarded after 8mok£uB the Low deL£veL-y cigarette were 1onset than r.he control c£KaretCe butts. These lonKer butts could be due to features of the desiKn of the ]Low dellvery ciKarette, much as the tlpp~nK paper lqch and vanC£laclon effects on the impact Ca £rrltat£ou rat£o. L£ttle difference can be seen in the way in which the motivational groups smoked "slchar the control oc the changed delivery cigarettes. The observation that the amount o£ compensation did not alter after a month of 8mokinK cigarettes with changed deliveries, sugKests that there £s unlike17 to have been phys£o~ogical adjustment Vithln the smoker duriu$ this period.. However the factors determ£n£ng the nlcotlne demand of a smoker merit further study. Most of the population are non-smokers up to m£ddle or Late taenaKe. When a person starts to smoke tt is usually a few cigarettes per week which rises to a £ew par day over a period of mouths. After the /~itiaL "experiments" with smok£r~K the p~rson settles at a level of ~0. 20, 40 per day and may sea7 at that Level £or years. L~ it is assumed that the ingluences on the number o£ cigarettes smoked par day age not entlgely £1nanclal the number o£ c/4BareCCes smoked per day (and the way they are smoked) m~sC be controlled by 4uternal mechanisms and some snvLroumental factors. One o£ chase may ba the rate at which nicotine is metabol£sed and/or excreted. Th£s, in turn, assumes that nlcotine Itse£f, and noc a metabolic derivative, is BAT Co LTD - MINNESOTA TOBACCO LITIGATION m ¢,..r,i %o w ¢,,..y'l

-30- the active principle in smoking. The faster the rate at vhich nlcocine is deactivated pharmacologically, the more £requencly the smoker must replenish his nlcocine "level" to provide the stimulation or ocher .revards 0£ n£coc£ne. Thls sussests that £C vould be appropriate co £nvesC£gace the meCabollc capabLllcy of indivlduals for n£coc£ne mud compare these dace v£th chel: moclvacions co smoke. REC0~ENDATZONS FOK TU~THEIt WORK ,, In order co make a better estimate 0£ the deliveries of a cigarette sacked by a human subject 4c 4s necessary Co reproduce the smokiug pattern of the smoker. Clearly £C is izpossib~e Co smoke the same clgarecCe twice, by macblne m~d human smoker, buc by smok£ng several s~n~lar c£garetces by machine to the human smoking pattern, and averas£ng the results, the chances o£ measurinz more accurate deliver£es £rom the cigarette are increased. Zc ~s intended thaC smok4ng patterns~ typical o£ an individual smoking partlcular brands, ~1 be dupllcaced and reported separately. Th~s rill involve sixteen subject:s, three phases o~ the exper/~ent and £ive duplicated repl£caces of the seZecCed ~2p£cal (average) smoking patterns i.e. 240 c£ga~ectes smoked by duplicator. The duplicated results wLll show TPM, nicotine and the carbon monoxide delivery of the gas phase. It ks also intended to duplicate all of the smoking patterns obta£ned £rom some individuals and thus to obcaln some idea oY the inter-subject vat, salons in del£vsrias. It was 8u4~Kested t:hat r.he est£uuttes o£ the numbers o£ cijarett~es smoked per day were noc accurate for the reasons already d£scussed. Zt ks hoped in future, Co obta£n a better esclmate of daily c£garetce consumption wlth an event recorder, vhich has already been descrLbtd (3). BAT Co LTD - MINNESOTA TOBACCO LITIGATION O g.-

-31- The results of the analysis o£ the filter t~ps for nicotine content showed that the estimated delivery of nicotine to the smokers was on ~vsrage about 1.6 mK grog the control ciSarecces. The average delivery . from the low delivery cigarette was about 1.5 ~ and from the high delivery cigarette about 2.1 ~ (the coe~£icient o£ v&r~ation for the data was about 4OZ). Taking 1.5-2.O mK of nicotine as a target value ft is suggested t~at this should be made available to the smoker with a minimum og T~H and 8as phase. Practically, he,waver, mnokers uould £ind a cigarette ~hich delivered the desired dose of nicotine i~ a small volume (e.g. 100 ml) o£ smoke generally unacceptable, as the high concentration o£ nicotine in the smoke vould result in a very high /space, poor sensory qualities and dL£flculty in controlling the self doslnS reg~m-.. An "ides1" design of cigarette might there[ore be one which vould provide a mnoker with 1.5-2.O ms o~ nicotine in about 350~450 ml of smoke, but at the sauna time deliver ~a low ~eight (5-10 ~) og TPH, end m lov concentration of c~cbon monoxide (1.O-2.0Z by volume). These 8u~estions may be compared vlth those made bY De. K.A.H. Russell (15) who has suggested that there is a case for a cigarette deL~verin$ 6 ms of tar TFH ratio o£ about 6. and 1.0-1.2 mg o£ nicotine i.e. a nicotine Cigarettes can be made to auccemd Dr. RusseLl's mpecificat4on by using tobaccos with. • nicotine content of about &.OZ and the design features shown in the Ciseretce Design Handbook (L4). Such cisaretces have bean made in limited quantities for experimental purposes (15). An alternative design strategy could be employmd but would ~equire the addition of nicotine compounds to a blend with low TPH potential. BAT Co LTD - MINNESOTA TOBACCO LITIGATION CO

-32- Such a blend m£sht contain a proportion of tobacco substitutes. This approach is, however, currently precluded by company policy. It is ztcc~anded chat cisaracCea ~r~ch TPM/nicotine ratios of spprox£mately 5:1 should be manufactured and studied £u a smoking behavlour experiment. Care should be taken in the fLnal desiKn of such cigarettes to optlmlse the positive attributes of the smok~nK qualities. Such design features ms draw resistance, tobacco aroma, flavour and irritation levels will need to be controlled as well as the deliveries of the c£aaretces. The design of this experiment, which has involved each subject mmokir~ the cigarette brands under investiaatlon exclusively for four weeks, with frequent laboratory measurements, has allowed a more accurate analysis of Ohm data. Thm analysis of these data have allowed us to present more c~aar cut answers than wars possible with the sin$1e shoc experimental designs used previously. It is therefore recommended thac a similar design should be used in future exper~encs to Investigate the differences in smoking patterns between brands. BAT Co LTD - MINNESOTA TOBACCO LITIGATION CO Co

-33- REFEKENCES l, Todd, G.F. Q Q 4. 5. 6. 7. e 9. 10. 11. 12. 13. 14. 15. Changes in the Smoking P&ttems in the U.E. Tobacco ~esearch Council, 1975. Health Departments of the Unlced ~dom "Tar and N£cot£ne YLelds o£ G~Ka=ettes". London D.E.$.S., 38nuary 1976. BAT Report No. RD.1300 Xestricted, 30.1.76. Kusse11, M.A.U. et a_..~l., Lancet, 1, 214, 1974. Russell, H.A.H. Nationvide BBC-TV, 23.5.75. BAT Report No. BD.1LO9-R, 7.6.74. P.usse11, H.A.H., Peto, 3. and Part1, U.&. 3.R. $tat£st. Soc. A. 137, part 3, 313-341, 1974. BAT Report No. RD. 1199-6, 28.4.75. BAT Report Ho. RD. 1007-2, 5.6.73. BAT Report No. L.458-R, 19.3.74. BAT Report Ho. RD.1322-Restricted, 30.1.76. Overton, 3. Ray. Beitrase zur Tabak~orschuug, 7, (3), 117-120, 1973. Passel1, M.A.~. Brit. Med. 3. 12th 3une 1976, paKe8 1430-1433. Randbooks of C£Sarette De8£~=, 3rd Ed£tion; 1976. Cou£erence On Smokiug BehavJ.ou]:, 11Ch and 12ch Occobez" 1976. BAT Co LTD - MINNESOTA TOBACCO LITIGATION ",,.C)

-34- APPENDIX STATISTTCAL EVALUATION OF DATA The £olLowlnK parameters have been considered in thl8 evaluaclon: 1. Number o£ puffs. 2. Butt Length. 3. Lit draw resistance (- total of pressure/coral readings to den puff duration). 4. Total time alight. 5. Tip nicotine. 6. Ear,hated nicotine. 7. Total of pressures co drav puff. 8. Average pressure. 9. Total volume. 10. Average volume. ].1. Total duration. 12. Average duracLon. 13. Total 4~cerva~. 14. Average interval. To simplify the procedure the experiment ~a ~malys,,d in tvo groups, the first (group l) iuvolvi.ag deem from all people r~ho mmoked the high delivery cigarette in phase 2; the second (group 2) involvinK data from all people who smoked the lov delivery cLgaretce in phase 2. A u~iva~ttte ulys:Ls of vcimnct has been perfo~ned for each parameter mud in uch Stoup using the follm,,iftg model for observations: YiJkl~ = U ÷ Di " $j ÷ Mk ÷ P(jk)l 4- DSIj ÷ DHik -,. SMjk "," DSMijk ÷ DPcjk)il ÷ ~(ijkl)m BAT Co LTD - MINNESOTA TOBACCO LITIGATION m k.,"l m

-35- where Y£jklm - the measured variable. E = the 8eneral mesh. D£ (£-1,3) - the ef£ecC due to de1£very (fixed). Sj (~=1,2) - the effect due to sex (f~ced). ~k (k=l,2) = the affect due co mocivat£on (fixed). P(jk)l (1=1,2) = She e£fecc due to person 1 with sex i m~d moc£vacion j (random). DS.. = the delivery-sex interaction (fixed). xj DMik = the dellvery-motivaCion interaction (fixed). SNjk = the sex-motivatlon interaction (fixed). DSH£jk = the del£verT-amc-motlvat£on interaction (fixed). DP(jk~L1 = the del£very-persou interaction (random). z(ijkl)m (m=l,lO) = the res£duaI error for replicate m of person ~ in the ijk comblnacion of the other factor levels (random). The three levels of the delivery factor correspond to the three phases of the exper~nenc. Alchou&h levels 1 and 3 actually involved ci&arectes rich th~ same dellvtry, they are create/ as separate delivery levels in the model to allow 8 Cast of the hypothesis that there is no d£fference in smokln~ beh~vlour for phase I mnd phase 3. The anslysls of variance table for the above model is shown £n Table II. A dot £nscead of a subscript ~apt£es thac a11 values of the G subscript have bean averased. Zt has been assumed that the error terms for replicates on i~d£vlduals arm random and have Independent normal distributions with zero mean and variance 02• The people effects and delLvery-person inceract£ons are also assumed Co be random, hav~nK i~dependenC normal d£strlbuclous w£ch zero means an~ varlances op2 and respectively. ODp BAT Co LTD - MINNESOTA TOBACCO LITIGATION J J

~llve~'y-9~rm |~rMtiea II 18L tld.m tepILr~ teJ 216 tm TeSaL -~(,- ~Tszs or v~tm,~ T~,n.z I~m ot Sq,,wres 30.2.2.2 ~ (TL .... . Y. oo.' )Z 1o.3.2.2 ~ ~r.j... - • ..... )z • o.3.z.: ~ {r..k.. " • ..... )t lO.:l jk~l f~.jkt. * ; ..... )t IO.Z.= ~j (•LJ... " 7f. .... " Y.J... * • ..... )z tO.Z.Z L (Y£.k.. " 7L .... " •..It.. * • ..... )z tO.3.2 ~ (7.jk*" * 7.j... " 7..I,. . * • ..... )z tO,2 l;k (Y|Jk.. " Y~J... " YL.k.. " Y.Jk,. • Yl .... * lr.j.,. * ]r,k.* " •,..~.)Z 10 i~kl (~lJkl. " •.Jkl. " YLJk., * •~ .... ): i~ktsl (]rLjk]Lm - yijlLZ.)"1 Ir, sc~,-,,tod ~ Square tI • rt • 10.~~t • ID.~DLI Sz , r1 , 30.%z • I~O.~Sjz Zs ,, r~ • ~m.epI ~ ZZO.~tk: Ik, - ~ * 30.st: s, - ,: • :o.~.~ - *o.~j.f~ • . - ~ • z*..~" - ~|k,,k g? - it: * Im.alp:~ - tO.]kJR~k |. " rt * 10.~| * 201~klMN|jk e~ BAT Co LTD - MINNESOTA TOBACCO LITIGATION

-37- The decision to reKard the people effect as random is open to criticism. It assumes that the Stoups of people used in the experiment (e-s. hishlY motivated males, lowly motivated males etc.) were selected 4n a random manner from the total populations of such people. Conclusions concerned with contrasts betweeu these groups will only be true for the parent populations to the extent that this assumption is correct. Since the panel was selected from volunteers within Group 6. & D. Centre 4t may be quite ~rcong to extrapolate to larger populations. The alternative procedure is to treat the people effect as £ix~d. This mesns that the subjects on the panel may only be treated as individuals. The results are then only strictly applicable to these individuals and, because the etatLstical tests become more sensitive, the danKer of makinl erroneous. extrapolations to larser populations may be increased. The ratios of mash sums of squares (F-r&tios) required to test hypotheses co~cern/~S tezw~ of the model maT be obtained from Table 11 by exan~n4nK the estimated m~sn squares. For emample, to test the hypothesis that there is no siKnlflcaut delivery effect on a variable (~D.2 = O) the ratio EI/E9 must be calculated for the variable. This & F-rat£o is then compared rich the appropriate percentaKe point on the F-d£stribucion and if the experinentei value is larKer, the hypothesis is rejected and there is sm~d to be a s£sni£icant difference between dellvecies at that level. For esch varLsble In the above llsc the appropriate Y-rat£os were calculated to test the sisnlfLcance o£ each o£ the terms ~n the above model (except, of course, the erzor term). A eu~nary of the s£gn4ftcant e£facts 4s presented in Tables 9 and 10 (Pass 27) where effects sL~ificant at 0.10 are sho~m by ~ and those 8isnLficant at 0.01 are shorn by ***. BAT Co LTD - MINNESOTA TOBACCO LITIGATION / (.J"l O~ m

-38- General Remarks on the InterprecaC£on of Tables 9 and i0 The oucsCandXng £sacure oE the results is cl~tc the variation betveen people is the major source 0£ variation in the data. lC is siKnifiean[ at the 0.01 level £or all v~b].es in both tables and is accompanied by very larks F-rac£os. There ia also a s£gu£flcanc dellvery/person ~nceraccion for the majority of varLables, buc the F-raClos are much smaller than for the m~t~n 'beL~een people' e££ecC. A s£Knificant dalivez~/person ~nCeracC£on £or a particular variable 4ndicates r~ac the ch&nKe in the value of the var£able £rom one delivery Co another is s£Knificantly dLfferenC for dif£erenc people. The larks variac£on between people vithin Kroups makes the tests £or di££erences beCveeu Kcoups o£ people less sensitive e-K. tests for dlfferences between ~he sexes and differences between the motivation Kroups. Thus, although few-s£Kn£ficanc sex and motivation e£feccs and their various ~nteractions were detected, £t may be that differences were missed because" the tests ware not sufficiently sensitive. circumstances permitted, this particular disadvanCmKe of the larks varlat£on baleen people may be offset by usinK more people per group to obtain better est£maCes of the Kroup me*,~s. Eovever, the main o~jective of the experiment is co test £or di£fecences in ~hs values of variables due to the dL~ferenc del4ver£es (phases). The tests for delivery ef£ects do noc have tb~s problem because the same people were used ~n each phase and the betveeu people variat4on is, therefore, not involved. S48nificanc del£very affects are observed in Tables 9 and 10 for many of the variables in both parts o£ the analys£s. Before d£scussing these affects a brief constant on motivational effects is appropriate. BAT Co LTD - MINNESOTA TOBACCO LITIGATION

-39- Motivation It can be seen £rom Table lO that moC£vatlon was only detected as a significant effect for three variables in group 2. The number of .puffs and the total pressure were both lover on average for the highly motivated group than for the lovly motivated group, while the average interval yes higher for the hiKh£y motivated group than for the lovly motivated group. These di££srences are the opposite of the expected di£fezences between those highly motivated Co smoke for nlcot£ne and those less hlghly aotlvtCed. Uovever, none of the di£ferences were 81~flcant at the 0.01 level and Ln group 1 (Table 9) no slgnific~nt ef£ect8 due to motlvacion vere observed whatsoever. The small 8~zs of the mot£vat;ion I~roups and the large "between persons vlth£n Stoups" variation may be distorting the obse:ved ef£ects due to motivation and concealing others. Ef£ect;s.,.og Chan6ed Dellvezy Raving establlshed by analysis of variance that delivery has a 81snificant ef£ect on a particular variable. £t is eonstruetlve to compare the mean values of the variable from each phase. The StudenClsed tanks test may be used to test the significance of di£ferences between the means. Th:ee differences between phases may be tested (1-3), (1-2). and (3-2). Since the cigarettes used in phases 1 and 3 vere the same, one would mcpect r3hat no s£Kn~tcanC differences between phases 1 and 3 would be observed and thac if there yes a significant difference bstveeu I and 2 there would be a similar difference between 3 and2. BAT Co LTD - MINNESOTA TOBACCO LITIGATION W

-4O- The Studentised ranKe test allowed the tst~sttim of least significant differences between me•ns. Differences Kr~tter than a least 8ilDificsmt dlZ£erence are then s•id to be siKnlflca~t. The results mre summarised in Tables 12 and 13 usinS an * to denote differences between mea~s silniflcant •t 0.1 and ** to denote differences s£gn~icant at 0.05. Xc can be 8ee~ from these tables that for group 1 there are only • few+ differences between phases 1 and 3 and the differences between 1 and 2 and between 3 and 2 •re frequently significant and similar Co each ocher. This is ~ch as expected. But for Kroup 2 the~e are more unexpected differences be~een phases 1 and 3, fewer siKn£ficant differences betveen 1 and 2 and in some instances the differences between 3 and 2 are not the same sign as between I and 2. A sugary of the statlst~c•lly significant differences due to changed delivery and •leo between the sexes ts presented £n Table 3 (pale 13). Effect of Time Zn order to test for a siKnif£cant carry over from one phase to the next or an accl/lat£sat£on process on mrltchLuK to a different delivery, the ten replicate maokinls £or each individu•l in each of the three phases were divided into three sub-phases, the first containing replicates i to 3, the seconds zepl£c•tes 4 to 6 and the third, replicates 7 to 9. Kepllcate 10 vas dropped. The nine new phases were than reKarded as 9 Isvel/ of a phue factor in a separate set of analyses of variance us:Lr~ the same model as before in •11 ocher respects. Th~s analysls •~Lowed the calculation of .least siKniflc•nc differences between levelsof the new phase factor and the ~esults o£ • comparison of means ace shown in Table 6 (pale 21) for ~roup 1 and Table 7 for rroup 2 skein usinl the * noCatlon. BAT Co LTD - MINNESOTA TOBACCO LITIGATION (3o O",

-41- TABLE 12 SIGNTFZCA/~. DIFFERENCES BETWEEN PHASES FOR GROUP I Variable 1-3 1-2 3-2 Number of Puffs Butt Lensch Total Pressure Drop Total Time Alisht TLp Nicotine Estimated Nicotine Total Pressure AvecaKe Pressure Total Volume Averase Volume Total Puff Duration Averase Puff Duration Total ~nCerval AveraBe Znterval ee te ~rA ** */t lit TABLE t3 SZGNTFIC~qT DrFFERENCES BETWEEN PHASES FOR GROUP 2 Var£able I-3 1-2 3-2 Number of Puffs ** Butt LenKth Total Pressure Drop Total Time Alisht Tip Nicotine ,r, Escinmted Nicotine ** Total Pressure ** AveraKe Pressure Total Volume ** Average Volume Total Puff Duration ** AveraKe Puff Duration Total Interval AveraKe Interval **- em ** e* **-- e.- e. e*- e.- e.- **- e- **- **- .&- (- denotes a negative dif£erence) BAT Co LTD - MINNESOTA TOBACCO LITIGATION m m

-42- The most consistent differences between levels are becveen levels 3 and 4 and between levels 6 and 7, correspond£nK Co the chtnSe co and from the different del£very cLsarette. Although ocher differences are present e.g. between 1 and 3 and between 7 and 9, no conslstenc differences in support of carry over from one phase co the next, or an acclimacisation period are observed. Zn particular, there 4s not a single sisnificanC difference between the first three repllcares and the last three replicates in phase 2 of the original experiment (levels 4 and 6) where an acclimaCisaCion effect: m£1hC have been expecced~. BAT Co LTD - MINNESOTA TOBACCO LITIGATION m OO

k=. o.#. ~) :.), • BAT Co LTD - MINNESOTA TOBACCO LITIGATION '~t2) Go)