Until techniques in biochemistry and high-resolution crystallographic electron microscopy were developed during the 1970s
which conclusively established the double helical shape and base-pairing rules for DNA, some scientists continued to question
the validity of Watson's and Crick's model. One of them was Jerry Donohue, the American physical chemist who in 1953
had alerted Watson to errors in the configuration of the atoms of the bases of DNA as represented in most textbooks of biochemistry,
enabling Watson to deduce the base pairing rule A=T and C=G. Donohue published an article in the September 12, 1969, issue
of Science (vol. 165, p. 1091), in which he questioned whether Fourier syntheses, a method for calculating the distribution
of electron density from amplitudes (seen in X-ray diffraction images) and assumed bond angles of molecules in crystals--and
thus a method for inferring the structure of such molecules--did in fact provide evidence for the Watson-Crick structure,
specifically the base pairs, as DNA researchers like Maurice Wilkins claimed. Donohue in his article presented electron density
maps of combinations of various base pairs which he claimed fit Fourier analysis just as well as the canonical A=T and C=G
pairs, thereby implying that other base pairs and thus other structures for DNA were conceivable, at least at the resolutions
then attainable in X-ray diffraction images.
Crick, along with Wilkins and other DNA scientists, refuted Donohue's criticism and arguments in letters published in
the March 27, 1970, issue of Science (vol. 167, p.1693). Donohue's letter was written in response to Crick's commentary,
in which Crick challenged Donohue to build models of DNA based on his purported recalculation of the X-ray evidence.
Number of Image Pages:
2 (171,829 Bytes)
1970-05-06 (May 6, 1970)
Original Repository: Wellcome Library for the History and Understanding of Medicine. Francis Harry Compton Crick Papers
Courtesy of the Wellcome Library for the History and Understanding of Medicine.
The National Library of Medicine's Profiles in Science
program has made every effort to secure proper permissions
for posting items on the web site. In this instance, however,
it has either not been possible to identify or contact the
current copyright owner.
If you have information regarding the copyright owner,
please contact us at
Medical Subject Headings (MeSH):
Embryology and the Organization of DNA in Higher Organisms, 1966-1976
Letter from Francis Crick to Jerry Donohue (May 20, 1970)
It is my understanding that in order for a group of reflections from an acentric structure to be effectively centric that
these reflections constitute a zone the axis of which is a twofold axis in the structure, such as the (h0l's in P21, or
the (hk0)'s, (h0l)'s, and (0kl)'s in P212121, or the (hhl)'s in P4m2. It would be easy to see if this were
not true for a DNA model if the table of F[calc] had been published, but unfortunately this is not the case.
Although there are numerous twofold axes in some models of one molecule these are not present in the structure as a whole,
a condition necessary for centric projections in certain directions.
But this is really a red herring. I took the stand that my reply should not cover every point in nauseous detail, otherwise
no one would read it. It simply is not true that "in centric structures my arguments have much less force". Numerous
errors have been made with centric Fouriers. To show this, however, would have required lengthy documentation which not only
would not have been appreciated by the general reader of SCIENCE but would have put him to sleep. Most of this was removed
from early drafts of my first paper on the advice of shrewd referees. Such as:
1. p-nitroaniline, Acta Cryst. 9, 960 (1956). This corrects an earlier error in which two incorrect centric projections had
led to a structure with absurd intermolecular distances. In the correct structure 21 out of 70 F's changed sign for one
of the projections. I haven't made the count for the other one. The two incorrect projections were considered plausible
by the authors - and the referees and the editor and numerous others who cited that paper.
2. triphenylene, Acta Cryst. 3, 165 (1950). The centric (hk0) projection, Fig. 2, was later shown to be incorrect in Acta
Cryst. 7, 595 (1954); see also Acta Cryst. 9, 173 (1956). 60 out of 105 F's changed sign in the correction. Many atoms
moved several A with respect to each other in the correction.
3. [beta]-selenium, Acta Cryst. 5, 236 (1952). In this determination two centric projections were correct, one was incorrect.
In the correction, Acta Cryst. 6, 71 (1953) one atom (of eight) was erased and manufactured elsewhere, and the intermolecular
relationships was grossly changed. The number of sign changes is not known because the first author did not publish an Fo
us Fc table for the projection in question. (I do wish authors would publish F tables.)
4. iso-leucine, Acta Cryst. 7, 703 (1954). Fig. 2a is correct, Fig. 1 is incorrect. They differ in the relative configurations
at C[alpha] and C[beta]. Out of 120 F's for this centric projection only eight differ in sign, and of these seven were
of medium weak intensity and one was of medium intensity!
There are many more examples in my notes, but I think that the above suffice to illustrate that it is easy to go wrong by
assuming that a particular model is correct and then "proving" it by calculating an electron density function on that
basis, whether or not signs or general phases are required. I saw no point in bringing this up in my reply to your comments,
but restricted myself to the dyad question.
Anne Sayre and someone named Flint have embarked on a biography of Rosalind Franklin. I suspect that they'll want your
version (if they haven't got it already).
P.S. Grandpa came from Cork, so note that I am an -ohue, not an -ahue.