2
Jon Marks Response to Britten De@rtments of Anthqology and Biology, Yale University, New Haven, CT 06520, V.S.A Carl W. Schmid Department of Chemistv, Universily of Califoornia, Davis, CA 95616, U.S.A Vincent M. Sarich Departments of Anthropology and Biochemistry, University of California, Berkely, CA 94720, U.S.A. Journal of Human Evolution (1989) 18, 165-166 Sibley & Ahlquist’s output was 3-fold: a tree, a set of distances, and a set of small standard errors associated with each distance. They have consistently maintained that the most appropriate analysis consists of treating the series of delta-T values across experiments as independent, and calculating mean distances and standard errors accordingly. We have shown that the standard errors published by Sibley & Ahlquist (1984, 1987) are not reflective of those inherent in their data when treated in this way; and they have now admitted publicly the existence of post-experimental numerical adjustments (Fellman, 1988; Lewin, 1988a,b). The falsification of the homoduplex control to experiment 1165, discussed by Britten, is but one of these. In their defense, Britten argues now that each experimental series should not be combined with others. We indeed indicated in a footnote the possible difficulty in pooling these data. Britten’s criticism should be leveled not at us, but at Sibley & Ahlquist, for we were simply trying to do what they had been advocating and claimed to have done. Is the melting temperature of the 1165 homoduplex indeed anomalously low, as Britten claims? In our Table 1, we show that it melted at 85.4 degrees. Yet this is but a tenth of a degree away from the control of 843 (Pun paniscus tracer), and 0.8 degrees higher than the other Homo sapiens homoduplex, experiment 115 l-l, the results of which were apparently not changed by Sibley & Ahlquist. We therefore fail to see any objective basis for Britten’s statement. Further, experiment 1165, whose H/G-H/C value of 0.77 Britten accepts, was apparently deemed so technically inadequate by Sibley & Ahlquist as to require numerical alteration; while experiment 1151, whose “exceptionally low” H/G-H/C value of 0.26 Britten eschews, was not considered by Sibley & Ahlquist to necessitate alteration. Thus Britten’s criteria for judging the quality of experimental results appear to conflict with Sibley & Ahlquist’s, and 0.26 degrees is actually within the range of polymorphic variation found in primates by Powell & Caccone (personal communication). The alteration of experimental controls, much less the modification of data in the absence of explicit criteria, should not be condoned. Review of this work has been 0047-2484/89/020165 + 02 $03.00/O 0 1989 Academic Press Limited

Response to Britten

Embed Size (px)

Citation preview

Jon Marks Response to Britten

De@rtments of Anthqology and Biology, Yale University, New Haven, CT 06520, V.S.A

Carl W. Schmid

Department of Chemistv, Universily of Califoornia, Davis, CA 95616, U.S.A

Vincent M. Sarich

Departments of Anthropology and Biochemistry, University of California, Berkely, CA 94720, U.S.A.

Journal of Human Evolution (1989) 18, 165-166

Sibley & Ahlquist’s output was 3-fold: a tree, a set of distances, and a set of small standard

errors associated with each distance. They have consistently maintained that the most

appropriate analysis consists of treating the series of delta-T values across experiments as

independent, and calculating mean distances and standard errors accordingly. We have

shown that the standard errors published by Sibley & Ahlquist (1984, 1987) are not

reflective of those inherent in their data when treated in this way; and they have now

admitted publicly the existence of post-experimental numerical adjustments (Fellman,

1988; Lewin, 1988a,b). The falsification of the homoduplex control to experiment 1165,

discussed by Britten, is but one of these.

In their defense, Britten argues now that each experimental series should not be

combined with others. We indeed indicated in a footnote the possible difficulty in pooling

these data. Britten’s criticism should be leveled not at us, but at Sibley & Ahlquist, for we

were simply trying to do what they had been advocating and claimed to have done.

Is the melting temperature of the 1165 homoduplex indeed anomalously low, as Britten

claims? In our Table 1, we show that it melted at 85.4 degrees. Yet this is but a tenth of a

degree away from the control of 843 (Pun paniscus tracer), and 0.8 degrees higher than the

other Homo sapiens homoduplex, experiment 115 l-l, the results of which were apparently

not changed by Sibley & Ahlquist. We therefore fail to see any objective basis for Britten’s

statement.

Further, experiment 1165, whose H/G-H/C value of 0.77 Britten accepts, was

apparently deemed so technically inadequate by Sibley & Ahlquist as to require numerical

alteration; while experiment 1151, whose “exceptionally low” H/G-H/C value of 0.26

Britten eschews, was not considered by Sibley & Ahlquist to necessitate alteration. Thus

Britten’s criteria for judging the quality of experimental results appear to conflict with

Sibley & Ahlquist’s, and 0.26 degrees is actually within the range of polymorphic variation

found in primates by Powell & Caccone (personal communication).

The alteration of experimental controls, much less the modification of data in the

absence of explicit criteria, should not be condoned. Review of this work has been

0047-2484/89/020165 + 02 $03.00/O 0 1989 Academic Press Limited

insuffkiently critical. and has consistently pcrmittcd results to bc published in spite of

methodological poverty. WC look forward to a time when the normal standards ofscientific

rigor will be applicable to DNA hybridization.

References