Smoking and Tobacco Control Monograph No. 7 
vary the test parameters and produce a range of reasonable smoking 
responses on say, tar, maybe using the machine and then to pick some 
number like the mean plus a standard deviation, let us say, just to throw 
something out for discussion, because this would represent a number that a 
reasonable number of smokers might really encounter. In other words, that 
would be the tar level that a substantial number of smokers would actually 
encounter in smoking a cigarette, and if that were presented to consumers, 
yes, it would err a little bit on the high side for some consumers, but 1 think 
our duty may be to give consumers information that serves to protect a 
reasonable number of those who are ingesting more. 
1 think that if that number were provided, 1 do not want to call it a 
maximum, you would find that firms would have an incentive to modify 
cigarettes. They have a lot of design features they can use to modify low- 
yield cigarettes to be sure that the mean plus one standard deviation would 
be as low as possible, and 1 do not think we should underestimate the 
importance of what is done here on the design of cigarettes in the future. 
1 do not think we should underestimate that, and I think if we give them 
something along the lines of what we are talking about, they have the ability 
to see that their cigarettes come in at as low a number as possible. 
DR. FREEMAN: May 1 ask you. Dr. Cohen, how would you reach the mean in 
such a method? 
DR. COHEN: 1 am certainly not technically competent, but in listening to 
the discussions and reading the papers, if the FTC testing method were 
adjusted to deal with such things as puff number, puff interval, and puff 
volume and this were done based on an observation of how smokers smoke, 
just as it was done when the original Cambridge Filter method was set up 
in the first place, then you would be able to know what the magic number 
would be for two-thirds of the sample or some arbitrary number, and it 
would be greater than the mean. I think that number would probably be 
a lot easier to communicate than a range, and it would have the side benefit 
of better informing smokers as to what their potential risk might be, and it 
would also provide great incentives to the industry to make cigarettes that 
came in at as low a number as possible. 
If one of the major problems with the low-tar cigarettes is where the filter 
holes are and how they work and the fact that they can be covered, then if 
this testing protocol were followed and the mean plus one standard deviation 
for that cigarette the way it was smoked were a fairly large range and if the 
company making that cigarette did not like that large a number, it has the 
capability of reducing that number by putting the filter holes in such a way 
that they are not going to be blocked. 
I would say that it is very important to consider the impact of what is 
done here on what they do. 
DR. FREEMAN: Dr. Shiftman? 
210 
