N 202

Birley’s study, all seven loci examined pro. ved to be under fairly intense selection These experiments do not show whethel the selection was acting directly on the lot concerned or on loci linked to then- (‘hitch-hiking effect’); but, irrespective 01 this, neutralist models cannot explain the speed at which allele frequencies changed It is, of course, possible that these result: are untypical or that the environment> used were inappropriate. Repetition of the experiments with a number of populationr and with environments simulating some of the fluctuations in environment ovel time or space found in nature should helf to decide this.

J.S. GAL1 Department of Genetics, University 01 Birmingham, Birmingham, U.K.

References Haigh, J. and Maynard Smith, J. (1972) Gener Res. 19, 73-89 Powell, J. R. (1971) Science 174, 1035-1036 McDonald, J. F. and Ayala. F. J. (1974) Nutun 250, 572-573 Minawa, A. and Birley, A.J. (1975) Narure 255 702-704

Limitations of mathematical models SIR: Milkman asserts, and Kimura agrees that the evidence from electrophoretic var. iation in natural populations makes the ‘infinite allele model’ untenable (see TIBS July 1976, p. N 152). This failure of ant of the earlier mathematical predictions o the neutral theory is due in part to the limited resolving power of electrophoresis and in part to improper simplifying assumptions of the model itself. The number of allelic states is not infinite [l and, as Kimura discusses, small negative effects of selection cannot be ignored especially in large populations [2,3]. Ye the question remains as to whether varia, tion on the molecular level is functiona (of value to the population), or is large]! noise due to mutation and chance. Kimur: and I hold the latter view.

Kimura and Milkman do not agree or the validity of threshhold models of mul, tiallelic selection, and here I side large]!

ERRATA

TIBS July 1976

Motoo Kimura - random genetic drift prevails: p. N 152, line 4 from bottom should read: ‘extremely low, broad-sense heritability (Hz = 0.004) [2]. This means that’

p. N 153, line 25 from bottom should read: ‘small number, preventing the value of n e becoming very large for’

with Milkman [4,5]. Except when it acts on severe mutants, selection appears to act primarily on polygenic traits, such as general health and disease resistance, and on individual loci in so far as they influence such traits.

Viability itself is a threshold trait in that an individual is alive until dead. Each genetic death, in this view, is due either to one severe mutation or to a whole lot of individually small effects; in the latter case, a given amount of selection can ser- vice many loci simultaneously. However, one comes again against insufficient mathematical models. In the case of trun- cation selection, the parameters of the model can be adjusted optimistically so as to allow for the maximum amount of selection per locus with the minimum amount of selection per individual; when this is done ‘substantial selection is easily possible at hundreds of loci at once’, as Milkman says. The difficulty is that there are at least 40,000 polypeptide specifying genes in the human genome [6]. About a quarter of these are thought to be electro- phoretically polymorphic, and a larger proportion must have non-electrophoretic

TIBS - September 1976

amino acid sequence polymorphism. The most optimistic selection model cannot apportion significant balancing selection to more than a small fraction of protein polymorphisms. In addition, some 99% of mammalian DNA does not directly code for protein, yet presumably includes gene- tic functions that are subject to natural selection.

As for allelic variation in Escherichiu coli, it is an excellent joke on those who would credit heterozygous advantage with maintaining very similar polymorphisms in diploid species.

JACK LESTER KING University of California, Santa Barbara, California, U.S.A.

References I King, J. L. (1974) Genetics 76, 607-613 2 Ohta, T. (1974) Nature 252,351-354 3 King, J. L. and Ohta, T. (1975) Generics 79, 681-

691 4 Milkman, R. D. (1967) Genetics 55,493-499 5 King, J. L. (1967) Genetics 55, 483-492 6 Bishop, J.O., Morton, J.G., Robash, M. and

Richardson, M. (1974) Nature 250, 199-204

Letter from Washington Nicholas Wade

Science court Does a pesticide cause cancer? Will the Pentagon’s latest piece of billion dollar hardware really work? Is the ozone layer at peril from aerosols? Scientists appear on both sides of the public debates about such issues, a matter of some disquiet to those who believe that the scientific method is so infallible, and the truths it arrives at so indisputable, that those who dissent must be either knaves or, what is worse, fools.

A proposal aimed at establishing unequivocally the scientific facts underlying political issues has been put forward by Arthur Kantrowitz, an expert on ballistic missile re-entry problems who is chairman of the Avco Everett Research Laboratory in Everett, Massachusetts. His idea is to set up a ‘science court’.

Experts appointed to argue each side of the case would attempt to agree on the scientific facts relevant to the issue. Those facts on which they could not reach agreement would be argued before the science court. The court, composed of scientist-judges, would then hand down a verdict as to what the scientific facts of the case were deemed to be. The court would

not touch any of the non-scientific aspects of the issue or make any public policy recommendations on what should be done about it.

Long ignored, the science court idea has recently been taken up by White House science advisory groups, of which Kan- trowitz is a member, and endorsed, somewhat grudgingly, by the National Academy of Sciences and National Science Foundation. Environmentalist Barry Commoner, on the other hand, has decried the science court as ‘a very serious attempt to reintroduce authoritarianism into science.’

The idea of a science court is attractive

The new ‘Science Court’