Intended for healthcare professionals


Publishing tobacco tar measurements on packets

BMJ 2004; 329 doi: (Published 07 October 2004) Cite this as: BMJ 2004;329:813
  1. Nigel Gray (nigel{at}, scientist,
  2. Peter Boyle, director
  1. International Agency for Research on Cancer, 150 Cours Albert Thomas, 69008 Lyon, France
  2. International Agency for Research on Cancer, 150 Cours Albert Thomas, 69008 Lyon, France

    Figures for tar, nicotine, and carbon monoxide are misleading and should be removed

    Admitting mistakes can be difficult, correcting them even harder. Labelling cigarette packets with tar yields (plus nicotine and carbon monoxide) was, and is, a mistake. The mistake was not in the conception of the low tar programme, or even in conducting it as a huge experiment with public health. The error was allowing the tobacco industry to control it.

    The tar delivery of cigarettes is routinely measured with a machine and, with the exception of the United States, stated on the packet as a legal requirement in almost every country in the world. It is accompanied by measurement of nicotine and often carbon monoxide.

    These measurements are now recognised to be misleading for two reasons, as is the simplistic concept of tar as a substance.1 2 w1 Firstly, human smoking patterns vary greatly and are not mimicked by the machine. Secondly, modern cigarette design facilitates compensatory smoking (over-inhalation), which may lead to the smoker taking in much greater amounts of tar and nicotine than are measured by the machine.3 The 1960s' word tar, often called total particulate matter, is a euphemism for what we now know is a chemical cocktail with at least 69 carcinogens and numerous toxins.4

    This practice has a long history and was originally legitimised by the US Federal Trade Commission,1 in an attempt to stop a “tar race” that had broken out between manufacturers. It was further supported by the public health establishment, which was swayed by evidence that tar painted on mouse skin gave a tumour dose response analogous to the dose response between cigarettes and lung cancer and implied that “the lower the tar and nicotine content of cigarette smoke, the less harmful would be the effect.”5 6

    This was a reasonable expectation in the context of the times, although the fundamental flaw in the concept was the lack of understanding of the dynamics of cigarette smoking and the extent to which they are driven by nicotine hunger. One did not expect that the tobacco industry would be devious or foolish enough to modify cigarette design in ways that made the modern cigarette at least as dangerous as its predecessor, despite a dramatic lowering of tar delivery.4 However, this was indeed what happened, and we now find the standard measurement allows the industry to fool both the system and the public.

    As well as facilitating compensatory smoking by the use of ventilated filters,7 other qualitative design changes led to increases in carcinogens,w2 specifically nitrosamines, which are plausibly involved in the well documented swing from squamous carcinoma to adenocarcinoma of the lung.8 This swing can only be attributed to changes in the composition of cigarettes. The concentrations of nitrosamines produced by cigarettes vary greatly both within and between brands and are controlled by manufacturers.9 10 This particular group of carcinogens should, and can, be greatly reduced or eliminated as a matter of urgency.

    Tar measurement and labelling has served the tobacco industry well. It has supported claims that cigarettes were light or ultralight and has seemingly, and falsely, reassured many smokers who might otherwise have quit the habit.11 w3

    If the measurements on the packet are misleading, is any measurement needed at all? Well, yes. Regulated upper limits need to be set for smoke carcinogens and toxins as they are for car exhausts.12 Some form of measurement is therefore needed for regulatory purposes, although not for labelling of packets, as no machine can mimic the variable habits of individual smokers.

    Canada has shown the way.w4 Manufacturers are required to submit reports on smoke emissions, under a testing system that eliminates filter ventilation by taping over the vents and raises puff volume from 35 ml to 55 ml, at puff intervals of 30 seconds instead of 60 seconds. This system is adequate for comparing and assessing brands for specified substances and, by eradicating the effect of filter ventilation on the smoking machine measure, may encourage the abolition of such filter ventilation. However, the Canadian government has not yet taken action to set upper limits based on what is reported and continues to require listing on the pack.

    It is now eight years since the US National Cancer Institute recognised the fallacies of the testing system, but we still have a paradoxical and unsatisfactory situation. Whereas the purpose for which the tobacco industry uses the measurements—underpinning descriptors such as light and mild—is under attack, the labelling of the packet with misleading figures for tar, nicotine, and carbon monoxide, is not. Descriptors such as light and mild have been banned in the European Union and are the subject of litigation in the United States, so we are likely to see the end of them in developed countries quite soon.

    The machine measured figures for tar, nicotine, and carbon monoxide should be removed from the packet, and a realistic measure must be established for regulatory purposes (as Canada has done). The current health warnings deal qualitatively with the risks of smoking very well, and misleading figures on the packet can only do harm.


    • Embedded ImageAdditional references w1-w4 are on

    • Competing interests None declared.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.