Product Design
Equations Relating Impact with Extractable Nicotine and Other Cigarette Variables
Abstract
Reports development of new equations relating cigarette impact with extractable nicotine, non-extractable nicotine and pressure drop [P.D.], says impact is an important aspect of sensory smoke panels and states this new equation makes it possible to predict "impact scores to within +/- 1.5 of the observed value 95 times out of 100." Indicates slight modifications of the smoking procedure were designed, notes it entails "extra work of a non-routine nature", and says these changes are not necessary solely for predicting impact. Lists results of 47 commercial cigarette brands representing 10 different markets and blend types, reports Total Particulate Matter [TPM], total nicotine and extractable nicotine and says the impact assessment panel smoked 47 "blind" brands along with a "full set of 9 impact standards". Reports deliveries [mg per puff] of: TPM, Extractable nicotine, Non-extractable nicotine, Total nicotine, P.D. and impact comparing standard vs. modified smoking. Discusses model equations relating impact to cigarette variables, explains how multiple regression analysis was used to derive the equations and summarizes the correlation matrices of variables with statistical analysis comparing extractable nicotine to pressure drop. States "Total nicotine, non-extractable nicotine and TPM are all correlated with impact, while pressure drop is inversly correlated as expected (the higher the pressure drop the smaller the puff, and thus the smaller the dose of smoke received).", and cautions low delivery cigarettes can give misleading results when the standard procedure is used.
Fields
- Author
- Hirji, Taj (BATCo GR&DC Process Scientist)Defense
- Wood, D. J. (BAT GR&DC Project Wheat)
Defense- Ayres, C. Ian (BATCO Group Research & Dev. Centre, UK)
Research Manager, BATCO GR&DC, United Kingdom, - Wood, D. J. (BAT GR&DC Project Wheat)
- Recipient
- Hughes, Ivor Wallace, Dr. (CEO Brown & Williamson, TI Executive Committee)Ivor Wallace Hughes was The Chief Executive Officer of Brown & Williamson Tobacco Company in 1983, also served on the Tobacco Institute Executive Committee in 1983 and was CTR Director 1/28/83.
- Sanford, Robert A. (BW RD&E VP, Research Director)
- Gibb, Robert "Rob" M. (ITC Attorney in Montreal CA)
Robert M. Gibb, Esq. was an attorney for Imperial Tobacco, LTD in Montreal, Canada, circa 1973-81- Wade, R. S. (BW R&D Director)
R. S. Wade is a former Research & Development director for Brown & Williamson Tobacco Corporation. Wade is deceased as of 1994 (G. Bell LT Waxman 5/16/94). Attorney for British-American Tobacco Co. Ltd. (1975-76); "R. S. Wade, Esq." received copy of "Project Wheat--Part 1" dated 7/10/75 and "Project Wheat--Part 2" dated 1/30/76 (Project Wheat 1&2).- Nicholls, R.G., Esq. (BAT Attorney 1975-76)
Nicholls was a attorney for British-American Tobacco Co. Ltd. in 1975-76. R. G. Nicholls, Esq. received copy of "Project Wheat-Part 1 dated 7/10/75 and Project Wheat-Part 2 dated 1/30/76 (Project Wheat 1&2).- Sottorf, H. E. (BATCo Scientist, Germany, 1975)
H. Sottorf was a Scientist at Batco in 1975. (Source: NM Tobacco Companies Personnel List)- Seehofer, F. (BATCo Scientist, Germany, c. 1975)
- Kruszynski, A.J. (Attorney for British-American Tobacco (BAT) Co. c. 1975-76)
Kruszynski was a attorney for the British-American Tobacco Co. Ltd. in 1975-76. A.J. Kruszynski, Esq., received a copy of Project Wheat-Part 1 dated 7/10/75 and Project Wheat-Part 2 dated 1/30/76 (Project Wheat 1&2).- de Siqueira, C.J.P. (Scientist at BAT 1975-76)
Scientist at British-American Tobacco Co. Ltd.(?)in 1975-76.- Felton, David Geoff Dr. (Sr. Scientist for BATCo R&D '75-76)
Dr. Felton was senior scientist for the Research and Development Dept of British-American Tobacco Co. Ltd. from 1975 to 1976. Felton received a copy of "Project Wheat--Part 1" dated 7/10/75 and "Project Wheat--Part 2" dated 1/30/76 (Project Wheat 1&2). - Sanford, Robert A. (BW RD&E VP, Research Director)
- Hypothesis
- Free Nicotine
- Design changes over time
Changes in cigarette design over the past half century.- Low-yield cigarettes
Modification of low yield products to assure that adequate levels of nicotine delivery are maintained, and effects of yield changes on toxicity and dependence.- Mainstream constituent yields
Modification of selected mainstream smoke constituents in response to health concerns.- Nicotine transport, transfer, and uptake
Design changes which alter nicotine delivery or effect how the product causes and maintains dependence, including transfer of nicotine from tobacco to smoke, and uptake into the body.- Sidestream constituent yields
Modification of selected sidestream smoke constituents in response to health concerns.- Sensory effects
Technologies used to measure, control, or alter sensory effects - Design changes over time
- Keyword
- Bound nicotine (Protonated or ionized nicotine)
- Brand differences
- Delivery modification
- Depth of inhalation
- Extractable nicotine
- Free nicotine (Unprotonated or unionized nicotine)
- High impact/low tar
- Impact (Throat grab)
- Inhalation (Smoke inhalation)
- Low delivery (Reduced delivery)
- Nicotine delivery (Smoke nicotine or nicotine yield)
- Per puff delivery
Per puff tar, per puff nicotine, and per puff CO- Puff parameters
- Sensory response
- Total particulate matter (TPM or Tar)
- Unextractable nicotine
- Cambridge filter pad
- Brand differences
- Smoke Constituent
- Gas phase nicotine
- Gas phase constituents
- Nicotine
- Total particulate matter
- Gas phase constituents
- Design Component
- Oriental tobacco (Turkish)
- Flue-cured tobacco
- Named Organization
- BAT Group Research and Development Centre
- *British American Tobacco Company Limited BAT (See British-American Tobacco Co.)
Defense- Brown & Williamson Tobacco Corporation (B&W)
Subsidiary of BAT U.S., located in Louisville, KY. - *British American Tobacco Company Limited BAT (See British-American Tobacco Co.)
- Subject
- Pressure Drop (Design)
- nicotine technology
- Bioavailability (Measures)
- Puff Parameters (Measures)
- Smoke Constituents
- Smoke Nicotine (Measures)
- Sensory Effects—Impact (Effects)
- Test/Consumer Preference (Testing)
- Test/Inhalation (Testing)
- Test/Smoke Condensate (Testing)
- Test/Smoke Constituents (Testing)
- Low Yield Cigarettes (Products)
- nicotine technology
- Brand
- Agrinion
- Belga Rouge
- Benson & Hedges (PM)
- BOULE DOR
- Boule Nationale
- Camel (RJR)
- CAPSTAN
- Carlton (ATC)
- Cavalla
- Chesterfield
- Churchmans No. 1
- CRAVEN A
- DUNHILL INTERNATIONAL
- Embajadores Suaves
- Embassy
- Embassy Extra Mild
- Embassy Ultra Mild
- ERNTE 23
- GALLIA
- Gaulois Caporal (Plain)
- Gauloise Caporal
- Gauloise Disque Bleu
- @gladstone_mild
- HB
Corporate Black- Izmir
- JOHNSON
- Kent (Lorillard)
- KRONE
- LORD EXTRA
- LUCKY TEN
- Marlboro (PM)
- MAROCAINE
- MARY LONG
- Monarch
- MURATTI AMBASSADOR
- NEW LOOK
- Pall Mall (ATC)
- Pall Mall Export
- PALL MALL EXTRA MILD
- PETER STUYVESANT
- PICCADILLY
- Players Mild N.C.
- Primeros
- PRINCE
- Raleigh (BW)
- Rothmans Ransom
- ROYALE
- Select
- SILK CUT
- SILK CUT EXTRA MILD
- St. Michel Vert
- State Express
- True (Lor)
- Vantage (RJR)
- Viceroy (bw)
- Winston (RJR)
- Belga Rouge
Document Images
Woo
~J
˘O
,--q
EQUATIONS RELATING IMPACT WITH EXTRACTABLE
-- NICOTINE AND OTHER CIGARETTE VARIABLES
REPORT NO. RD.1337 RESTRICTED
22.3.1976
Group .
ReseaPcn
a#dllevelopme#t
/:entre
SOUTHAMPTON ENGLArq E;=
BRITISH- AMERICAN TOBACCO COMPANY LIMITED
Brown ~: Wiili~Imson
APR 6 1976
RESEARCI-I LIBRAR Y
~?0311988
This confidential report is the property of Britl.gt; Amertcan Tohaccc~ Compaiw Limited. and must
not t)e copmd or sh;~wn tn uf~a~Hr~,, ......... )erso~,'.
EQUATIONS RELATING IMPACT WITH EXTRACTABLE
NICOTINE AND OTHER CIGARETTE VARIABLES
REPORT NO. RD.1337 RESTRICTED
22.3.1976
AUTHORS: D.J. Wood
T. Hirj i
ISSUED BY: C.I. Ayres
PROJECT JOB NO: 175
DISTRIBUTION:
Dr. S.J. Green
Dr. I.W. Hughes
Dr. R.A. Sanford
R.M. Gibb, Esq.
R.S. Wade, Esq.
R.G. Nicholls, Esq.
Herr H. Sottorf
Dr. F. Seehofer
A.J. Kruszynski, Esq.
Dr. C.J.P. de Siqueira
Dr. D.G. Felton
Library
File No. 46E
Copy No. I, 2, 3, 4, 5
" " 6
" " 7, 8
tl tt 9
" " IO, Ii, 12
" " 13, 14
" " 15
" " 16
" " 17
" " 18
" " 19
" 20, 21
" " 22
COPY NO:
fi. 7031i989
DJW/TH/LK/46E
Group Research & Development Centre,
British-American Tobacco Co. Ltd.,
SOUTHAMPTON.
22nd March, 1976
EQUATIONS RELATING IMPACT WITH EXTRACTABLE
NICOTINE AND OTHER CIGARETTE VARIABLES
(Report No. RD.1337 Restricted)
SUMMARY AND RECOMMENDATIONS ,;~
Impact is one of the sensations that sensory smoke panels are
trained to assess. Equations have now been derived relating impact
scores (on the "new" scale of impact) with extractable nicotine,
non-extractable nicotine and pressure drop. The preferred equation,
based on standard P.D.L. smoke analysis, has a correlation coefficient
of O.918. Using this equation it is possible to predict impact scores
to within ±1.5 of the observed value 95 times out of I00. A slightly
improved prediction results from the use of a modified smoking procedure
which is designed to optimise conditions prior to the measurement of
extractable nicotine. However the modified procedure entails extra work
of a non-routine nature, and its adoption is not at present justified
when extractable nicotine values are required solely for predicting
impact. But when extractable nicotine values are used to give information
about blend type the modified smoking procedure is recommended, because
in the case of low delivery cigarettes use of the standard procedure can
give misleading results.
570311990
--2--
INTRODUCTION
A previous report on this subject (I) sought to explain the scores
for impact given by a trained sensory smoke panel to a range of cigarettes
in terms of extractable nicotine, non-extractable nicotine and pressure
drop. As a result of a stepwise multiple regression procedure an equation
was derived with a multiple correlation coefficient of O.913; this equation
explained 83.4% of the variation about the mean. As pointed out in that
report, the manner in which the impact scores were obtained suffered from
various limitations:
(i) Results from a number of separate exercises were combined;
the panel inter-compared cigarettes within each exercise, but
not between the various exercises.
(2) Within the time span covered by these exercises (approximately
3 years) the composition of the panel inevitably changed.
(3) The "old" rather than the "new" impact scale was used.
It was felt that the precision with which impact scores are obtained
could well be improved by removing the above limitations.
With regard to extractable nicotine, which was the most important
single term in the regression equation, this is normally determined as
a result of smoking batches of 5 cigarettes, the smoke from each batch
of 5 being trapped on a single Cambridge filter pad. Since the smoke
deposited on the pad is made up to a standard volume (50 ml) with water
prior to extracting with chloroform, irrespective of the delivery per
cigarette, the concentration of smoke in the aqueous solution can vary
within wide limits. It has been shown previously (2) that the proportion
of nicotine which is extracted by chloroform depends on the pH of the
570311991
-3-
aqueous solution, and that the latter depends on the concentration of
smoke in the aqueous solution: for low delivery cigarettes attainment of
equilibrium pH necessitates dissolving the smoke from more than the
standard 5 cigarettes in 50 ml water. So there could be a case for
modifying the method used for determining extractable nicotine.
The objectives of the investigation now reported were:
(I) To derive an equation relating impact (measured on the "new"
scale) with extractable nicotine and other cigarette variables,
after taking steps to improve the precision of impact assessment.
(2) To decide whether any change is justified in the method for
determining extractable nicotine.~
ASSESSMENT OF IMPACT, DETERMINATION OF DELIVERIES
For the purpose of deriving equations, 47 commercial cigarette
brands were chosen. Between them these brands represented I0 different
markets, blend types ranging from flue-cured to dark air-cured, and a
wide range of impact intensities.
For impact assessment the brands were presented "blind" (i.e. with
brand names obscured), usually two at a time, to a 6-man panel, along
with a full set of 9 impact standards (3). The same six members assessed
all the brands. Each brand was smoked by the panel on at least two
occasions, and further assessments were carried out where necessary until
it was judged that a dependable impact score had been obtained, judgement
being based on the agreement between repeats and the spread of individual
results.
For the determination of TPM, total nicotine and extractable nicotine
deliveries the cigarettes were smoked using two different procedures:
--4--
(I) Standard Procedure - Four batches of 5 cigarettes were smoked,
each batch of 5 being smoked onto a single Cambridge filter
pad. TPM, total nicotine and extractable nicotine trapped by
the pads were determined in the usual way.
(2) Modified Procedure - From the results of the standard procedure,
the number of cigarettes (to the nearest whole number) required
to deliver approximately i0 mg total nicotine per Cambridge pad
was calculated. The calculated number of cigarettes was smoked
as before in quadruplicate (or in duplicate where the required
number of cigarettes exceeded 15 per Cambridge pad), and the
TPM, total nicotine and extractable nicotine trapped on each
pad were determined in the usual way. For six of the brands,
smoking by the standard procedure delivered approximately I0 mg
nicotine per pad, so the modified procedure was not necessary.
The deliveries (in mg per puff) resulting from the standard procedure
are shown in Table i; those resulting from the modified procedure are shown
in Table 2. In these tables non-extractable nicotine is the difference
between total nicotine and extractable nicotine.
After equations based on the original 47 brands had been derived, a
further 23 cigarette samples were smoked by the standard procedure to
see how well the impact predicted by the appropriate equations agreed
with the impact as assessed by the panel. The samples consisted of 20
commercial brands and 3 samples representing different types of Oriental
tobacco. The results are given in Table 3. It will be noticed that
some of the brands in Tables I and 2 appear again in Table 3, but as they
were sampled at different times they were not necessarily manufactured to
the same specification.
"" 5703119
--5--
DISCUSSION
In attempting to derive equations relating impact with certain
cigarette variables, two basic models were used:
A. Impact = a + B (dose)
B. Impact = ~ ÷ ~ log (dose)
Model A is of the type considered in the earlier report (I). Model B is
an example of the linear response, log (dose) relationship first proposed
by Fechner (4). In each case the expression "dose" represents some
combination of the cigarette variables that were measured; these combinations
were not only of the simple additive type, but included al~b the ratio
between one variable and another, the product of one variable and another,
and so on.
Stepwise Multiple Regression analysis was used in deriving the
equations, as explained previously (I). Table 4 shows the correlation
matrices for the untransformed variables (i.e. Model A) and the log
variables (i.e. Model B) relating to the standard smoking procedure;
Table 5 shows the same data relating to the modified smoking procedure.
To avoid making the matrices too cumbersome, cross product terms,
ratios etc. are not shown, although they were of course included in the
analysis. From the matrices it is seen that the single variable most
highly correlated with impact is extractable nicotine, the correlation
being somewhat higher for the log deliveries than for the untransformed
deliveries. Total nicotine, non-extractable nicotine and TPM are all
positively correlated with impact, while pressure drop is inversely
correlated as expected (the higher the pressure drop the smaller the
puff, and thus the smaller tile dose of smoke received).
, 70311994
-6-
~r
i" 4
Using the stepwise procedure four equations were derived:
Standard smoking, untransformed deliveries
Impact = 1.8050 + 59.7038 (ext. nic.)
+ 9.7178 (non-ext. nic.) - 0.1153 (P.D.) .....
(S)
!
R = 0.9180 R2 = 0.8427
Standard smoking, log deliveries
Impact = 16.5619 + 2.5572 loge (ext. nic.)
- 1.7821 loge (P.D.) .....
($2)
R = 0.8925 R2 = 0.7966
Modified smoking, untransformed deliveries '~
Impact = 2.1449 + 52.2070 (ext. nic.)
+ 10.6666 (non-ext. nic.) - O.1143 (P.D.) .....
(M)
1
R = 0.9265 R2 = 0.8584
Modified smoking, logdeliveries
Impact = 14.O213 + 1.7372 loge (Ext. nic.)
- 1.6967 loge (P.D.) .....
(M)
2
R = 0.9080 R2 = 0.8246
In these equations the deliveries of extractable nicotine and non-
extractable nicotine are in mg per puff, pressure drop is in cm W.G.;
R is the multiple correlation coefficient, R2 is the proportion of the variation
about the mean explained by the regression.
Summaries of the respective analyses are shown in Tables 6-9.
Tables i0 and Ii give impact values predicted by each of the 4 equations
for the 47 brands, together with the 95% confidence limits and prediction
limits for these values. The 95% confidence limits refer to the probability
that, based on the measured values for the samples examined in this
investigation, the true mean lies within the limits given.
The 95%
S?0311995
^
C
prediction limits refer to the impact score which the panel is likely to
give as a result of smoking a further sample, not necessarily of the same
brand, with identical smoke deliveries and pressure drop.
Plots of observed impact against impact predicted by each of the
four equations are shown in Figures 1-4. In Figure 2 the points are
rather widely scattered, particularly at low values of impact. The
scatter is less in Figures i, 3 and 4; visual inspection suggests that
there is little to choose between these three regressions, and this is
confirmed by the correlation coefficients which have values of O.918,
0.926 and 0.908 respectively. Two of these represent a slight improvement
over the correlation coefficient of O.913 in the previous report (i) in
spite of the fact that the equations are very much simpler than the
somewhat cumbersome equation derived previously.
Table 12 shows values for impact predicted by equations S and S
1 2
respectively for 23 additional brands. Since the data on these brands
were not involved in deriving the equations they form a useful means
of testing the predictions. From Table I0 it is seen that the 95%
prediction limits are around ±1.5 for equation S and around ±1.7 for
i
equation S . In Table 12 the difference between observed and predicted
2
impact falls within these limits for all samples except two, the two
exceptions both being cigarettes containing 100% oriental tobacco.
Such cigarettes did not feature among the original 47 samples, and it
is possible that a detailed examination of the smoke from these two
tobaccos might reveal some peculiarity.
Comparing the correlation coefficients for the four equations it
appears that changing to the modified smoking procedure described
L~
.k
-8-
earlier in the report would result in a slight improvement in the accuracy
with which impact can be predicted. However inpractical terms this
slight improvement is not sufficient to justify the considerable amount
of extra work that the modified smoking procedure entails. It is true that,
in the case of the low delivery cigarettes included in this investigation,
the modified smoking procedure led to a much smaller proportion of the
nicotine being extracted into chloroform, a result that was expected from
previous observations (2). It is also evident that the replication of
total nicotine determination and of TPM determination between the standard
and modified smoking procedures was in many cases far from good (compare
Tables I and 2) although there is no reason why these determinations should
be affected by the change in procedure. The results do not show whether
it is the inherent cigarette variability or experimental error that is
responsible; but given this general lack of precision in determining
smoke deliveries it is perhaps understandable why attention to one
particular step, affecting the proportion of nicotine extracted by
chloroform, should lead to only limited improvement in the correlation
between observed and predicted impact. Of the two equations based on
standard smoking, equation S gives the closer prediction and is to be I
preferred.
There is one area of work, unconnecte4 with the present, where
a change to the modified smoking procedure is justified. When examining
competition products the percentage of nicotine that is in the extractable
form can be a useful indication of the type of tobacco in the blend.
However unless steps are taken to smoke sufficient cigarettes per
Cambridge pad for pH equilibrium to be reached in the aqueous smoke
