Anecdotal, Historical and Critical Commentaries on Genetics
Edited by James F. Crow and William F. Dove
EVEN if R. A. Fisher's elucidation of the human blood-group system Rhesus in terms of the three linked loci C, D, and E had not proved to be substantially correct, it would still have been an outstanding example of the power of analytical thought to unravel a complex array of genetical data. In fact, as a recent review relates (Avent et al. 2006), D is one gene (carrying the D antigen) and C and E are different splicing forms of another (CE carrying the C or c antigens and the E or e antigens). Fisher's solution is recognizable beneath the modern molecular detail.
The story of the unravelling of the Rhesus puzzle is told in chapter 13 of Joan Fisher Box's R. A. Fisher: The Life of a Scientist (1978). As well as describing the serology, Box gives a comprehensive description of the wartime circumstances, which in 1943 reunited in Cambridge Fisher and his former London colleague R. R. Race. Box was able to draw on Race's unpublished 1968 Fisher Memorial Lecture “Blood Groups in Human Genetics” in which he described the meetings with Fisher in The Bun Shop, a Cambridge public house, where the interpretation of the Rhesus reactions was discussed over pints of beer. Much later, he and his wife Ruth Sanger told the story (Race and Sanger 1982), and on the occasion of the centenary of Fisher's birth it was repeated by Clarke (1990) and Bodmer (1990), both of whom also knew Fisher. Bodmer (1992) gave a fuller account still to the Eighth Congress of Human Genetics in 1991. Fisher himself described the purely scientific development in a lecture that he gave at Woods Hole, Massachusetts, in 1946: “The Rhesus Factor: A Study in Scientific Method” (Fisher 1947).
The purpose of this article is simply to bring together in one place for the first time five illustrations that belong to the story, to remark briefly on Fisher's combinatorial skills and his relations with A. S. Wiener, and to add a personal coda. It is best to take first the diagram (Figure 1) from Fisher (1947) because it explains the relationship between the original Rhesus notation and that proposed by Fisher. As Fisher writes, “We may represent the eight heritable antigen complexes geometrically as the corners of a cube, while the six elementary antigens are represented by the faces; each allelomorphic pair of antigens is then a pair of opposite faces, and the three faces meeting in any point specify the antigens in each complex.” (Ry) and (CdE) are in parentheses because the anti-d antibody had not yet been discovered (nor has it to this date); it was part of the brilliance of the hypothesis that this haplotype and some of the missing antibodies that the model predicted were later found. This is one of the very few diagrams in the whole of Fisher's work, surprising in view of Fisher's geometric way of thinking.
The two notations for the Rhesus antigens from Fisher (1947).
Next is Fisher's solution as he wrote it out on the back of a piece of Caius College notepaper (Figure 2), probably after dinner in College (where he lived) on the day that Race had told him of his latest results when they met in The Bun Shop. It was presented to the Fisher Memorial Trust by J. J. van Loghem and is reproduced with the permission of the Trust. The “CDE” notation has not yet put in an appearance, but the tables may be readily interpreted by reference to Figure 1. Note in particular the two faces of a cube at the top of Figure 2 and the list of some of the genotypes and their reactions on the right, divided into “5 common,” “5 rare,” and “2 not observed.” Of this occasion Race and Sanger (1982) wrote, “The immediate reaction was puzzlement and lack of understanding, especially on the part of Race; colours and even music were tried but did not help; eventually unaided understanding alone came to the rescue, and thereafter it was impossible to believe anything else.”
Fisher's 1943 solution to the Rhesus complex.
Figure 3 is the beer-stained piece of paper on which, in The Bun Shop the following year, Fisher outlined his crossing-over hypothesis explaining the occurrence of the rare antigens. He uses the CDE notation. The piece of paper was presented to the Fisher Memorial Trust by Ruth Sanger with the annotation, “This was the first writing down by Professor Sir Ronald Fisher of his very elegant idea that the less frequent Rh chromosomes might have arisen by crossing-over in heterozygotes for [sic] the more frequent chromosomes. He wrote it in ‘The Bun Shop,’ a ‘pub’ in Cambridge, on 22nd June 1944. The Professor is very short sighted and was not aware of a good deal of beer on the table—the cause of the marks on the lower part of the paper.” The figure is reproduced with the permission of the Trust.
Fisher's 1944 explanation of the rare antigens.
In the telling of the story, the two sessions in The Bun Shop have tended to be confused or at least conflated (e.g., Clarke 1990) but the evidence is that the first was in late 1943 (as Race once said) and the second was indeed in the summer of 1944. The date of the latter is significant, for Race's letter to Nature, “An ‘incomplete’ antibody in human serum,” was published on June 24 (Race 1944). With his characteristic generosity toward his students (Race was registered for a Cambridge Ph.D.) Fisher did not co-author the letter in which Race wrote, “The research arose out of a suggestion by Professor R. A. Fisher,” adding, “The three forms of allelomorphic antigens are arbitrarily denoted by C, c, D, d, E, e, chosen to avoid confusion with any symbols so far used” (presumably the allusion is to A and B of the ABO system); the letter is dated May 17 and Race would of course have seen the proofs of it by the time of the meeting in The Bun Shop on June 22. Fisher and Race (1946) record that the notation was Fisher's idea, and Taylor and Race (1944) wrote, “The terminology used in the present account of the Rh system [that is, the original terminology] will certainly not be permanent. Race (1944) has described a most ingenious and attractive scheme, formulated by R. A. Fisher, for the Rh genes, antigens and antibodies.”
Race submitted his Ph.D. thesis “The Rh Blood Groups” (Race 1948) on March 24, 1948. It should be a valuable source for any historian of serology. (On examining the copy in Cambridge University Library, I find that I am the only reader ever to have consulted it.)
Fisher noted that cDe could be produced by three different crossovers among the three commoner types and that this was more common than the three rarer types. From this, he suggested the chromosomal sequence DCE, with C and E being close together and separated from D by a longer distance (Box 1978, p. 365). At the time, the evidence was not very strong, but it now looks as if, once more, Fisher was on the right track because, as mentioned earlier, it is now known that D is one locus, whereas C and E are alternative splice forms of a single gene.
Next is a picture of The Bun Shop itself (Figure 4), kindly supplied by Sir Walter Bodmer and already published in Clarke (1990). Sir Walter and I joined the Cambridge Department of Genetics at the start of Fisher's last academic year as professor (1956–1957) and we both remember visiting The Bun Shop with Fisher after his elementary lectures the previous academic year. It was pulled down not long afterward during the redevelopment of that part of Cambridge. (The redevelopment has itself since been pulled down and the site is currently undergoing another cycle of building; visitors to Cambridge should not be misled by the fact that there is now another Bun Shop elsewhere in the city.)
The Bun Shop public house in Cambridge.
Because Fisher was reelected to a Fellowship at his old college, Gonville and Caius, on assuming the University's Arthur Balfour Professorship of Genetics in 1943, he supplied a fresh photograph for the Fellows Photograph Album (Figure 5), contemporary with the elucidation of Rhesus. It is reproduced by courtesy of the Master and Fellows.
R. A. Fisher about 1943.
THE COMBINATORIAL BACKGROUND
That Fisher was an ace combinatorialist is obvious from even a cursory examination of his writings. His memorial window in the Hall of Caius College indeed reflects this by depicting a Latin Square (Figure 6, from the dust jacket of his book The Design of Experiments; Fisher 1935). This talent was deployed to great effect in the unravelling of genetic systems. In section 16 of his path-breaking paper, “On the correlation between relatives on the supposition of Mendelian inheritance,” Fisher (1918) considered the effect of “coupling” between two loci (that is, linkage) and, in particular, the two extreme cases of no linkage and complete linkage. “The above analysis of polymorphic factors enables us to compare these two extreme cases; for there are 9 phase combinations of a pair of dimorphic factors, or, if we separate the two kinds of double heterozygote, 10, which … can be interpreted as the 4 homozygous and the 6 heterozygous phases of a tetramorphic factor.” In their commentary Moran and Smith (1966) write, “This mapping of a system with two factors at each of two loci on to a system with four factors at a single locus is particularly interesting and can be illustrated as follows.” The reader is referred to Moran and Smith for the further details.
Commemorative window to Fellows in Caius College, Cambridge, UK. Clockwise from top left: C. S. Sherrington, John Venn, George Green, James Chadwick, R. A. Fisher, and F. H. C. Crick.
Here we have, of course, the two-locus precursor of the three-locus CDE scheme, in which the genetical situation can be mapped either onto a single multiallelic locus (A. S. Wiener's preferred solution; see below) or onto three fully linked diallelic loci (Fisher's solution). Fisher was keen to learn of any crossovers that would have confirmed a physical basis for his hypothesis, but even without them the three-locus analysis offers an enormous clarification of thought by comparison with its rival, as Race observed (“thereafter it was impossible to believe anything else”).
Fisher will of course have later been perfectly familiar with F. Bernstein's 1924 conclusion that the ABO blood-group reactions could better be explained by a single triallelic locus than by two unlinked diallelic loci, the reverse of the Rhesus hypothesis. The grounds in this case were the population frequencies of the genotypes. The most accessible reference for this is the worked analysis in my book Likelihood (Edwards 1992) where, incidentally, a footnote acknowledges the help of R. R. Race with Bernstein's original papers.
Bodmer (1990, 1992) recollects that Fisher told him that his Rhesus hypothesis was also influenced “by the Swiss geneticist Ernst's interpretation of the Primula incompatibility system as a complex of closely linked loci,” as confirmed in a letter Fisher wrote to Ernst in 1957 (Bennett 1983): “It is a pleasure to send you one of my remaining copies of the paper I gave at Woods Hole in 1946 [Fisher 1947]. I was, indeed, influenced in forming my ideas about the Rhesus complex by the system you had first proposed for the factor in Primula determining dimorphism.” A reference to A. Ernst's 1933 article may be found in Bodmer (1960). There were still some copies of Fisher's article in his offprint boxes in 1957. I possess one of them.
A. S. Wiener:
Wiener, one of the co-discoverers of the Rhesus system, was a strong advocate of the original notation, which portrayed a single multiallelic locus. He was never reconciled to the CDE notation, even after this had become the norm. A prolific writer, he took every opportunity to criticize it. Fisher, although of course always recording the role of Wiener as one of the discoverers of the system, limited himself to oblique references such as, “The efforts to force a notation on the system which does not recognize its genetic character, has been, and perhaps still is, a source of controversy” (BBC radio talk, July 17, 1958; Fisher 1958). I am sorry to have to report that when Wiener started to give his invited article to the Blood Groups session of the Second International Conference of Human Genetics in Rome in 1961 (Wiener 1961), we witnessed Fisher, who seemed to have occupied a front seat deliberately, rise from his seat and, slowly picking up his hat and stick, ostentatiously walk out.
In earlier times Wiener and Fisher probably had a friendly relationship. There are articles co-authored by Wiener in the Annals of Eugenics during Fisher's editorship. One entire box in Fisher's offprint collection is devoted to Wiener, and several of the articles that Wiener sent are autographed. Of particular interest to me is “Method of measuring linkage in human genetics” (Wiener 1932), autographed “To R. A. Fisher, With sincere appreciation,” an important paper that I missed in my history of linkage estimation (Edwards 2005). There is even a typescript of “Genetic theory of the Rh blood types” later published as Wiener (1943), which appears to be the copy to which Race et al. (1944) refer when they say, “This was the state of our work when a letter, dated October 11, 1943, came from Wiener enclosing the typescript of a paper then in the press, in which is described the behaviour of six allelomorphs of the Rh gene.” This typescript was certainly seen by Fisher because on the back is a note in his hand giving details of a mating in mouse genetics taken from Grüneberg (1943), The Genetics of the Mouse (Mary Lyon kindly identified this for me).
The arrival of this typescript and the Cambridge discovery of the fourth antiserum (Race and Sanger 1982) were the stimuli for Fisher's “Bun Shop” proposal of the three-locus complex, which must indeed have taken place some time after mid-October 1943.
A personal coda:
As a graduate student I was sent by Fisher to Race and Sanger's laboratory, by then in London again, to learn simple blood-grouping techniques, and I kept in touch with them thereafter. As Race mentions in his Fisher Memorial Lecture, he was also in touch with J. H. Edwards, my brother, both in connection with one of John's patients with an abnormal chromosome 18, and his computer program for determining the origin of the X chromosomes in Klinefelter's syndrome. Many years later John and I published a joint article (Edwards and Edwards 1992), which happily reflected these links with the Races and can serve as a coda to this story.
We showed how the Rhesus frequencies published by Fisher and Race (1946) could be represented in a square diagram in which the area designated for each haplotype accurately reflects the proportions that would obtain if there were no pairwise associations or threefold interactions between the loci. When the haplotype frequencies are added in the form of dots randomly placed in the corresponding areas, the departures from the equilibrium state are readily perceived (Figure 7). “The diagram clearly displays what Fisher himself noted, that the chromosomes are not in linkage equilibrium and that they fall into three groups, common, rare, and absent (i.e., extremely rare). The extremely rare chromosome, CdE, is entirely surrounded by three rare chromosomes, Cde, cdE, and CDE, these being the only three from which it could be derived by the mutation of a single gene…If CdE is deleterious but is maintained by mutation this may explain its rarity, though Fisher actually suggested crossing-over rather than mutation.”
(a) The Rhesus haplotype frequencies (see the text for explanation). (b) The key to a.
Note:
The account related here involves only Fisher and Race in the discussions in the Bun Shop in 1943 and 1944. It is often assumed that Ruth Sanger was there, too, but she is not recorded as having arrived from Australia before 1946. The impression that she was present can easily arise from the use of “we” by Race and Sanger (1982), but this can be interpreted as meaning Race and a colleague or colleagues from the Blood Group Laboratory, perhaps G. L. Taylor. I should be glad to be corrected on this point.
- Copyright © 2007 by the Genetics Society of America
