Вы находитесь на странице: 1из 2

July 25, 19631

is no obvious formula for its construction ; and what is obvious, of course, varies between observers.




surprise that present probability theory applied

and if 'we

science in fact works as well as it does ;








could discover how it is that it does work so well, are might be able t,o see how it is t,hat it does not appear to work quite well enough. I wish to express my grat'itude to t'he Perrott Electors of Trinity College, Cambridge, under whom the new description of this psychical research problem was originally made, and to Prof. A. C. Hardy, of

Oxford, who generously provided a grant for the

research to cont,inue in his Department. I should

like to t,hank Sir Ronald Fisher, by whose writings and

conversation I have been greatly helped.


' Coover,

John E.,



Psychiral Research"

University Press, 1917).

' Thoules~i,R. H., Proc. Soc. Psych. Res., 43,21 (1935). "hine, J. B., "Extra-Sensory Perception" (I.ondon, 1935):

Vorwald, R ., J.

"Richmond, Nigel, J. Soc. Psych. Res.,36,577 (1952).

.Purupsychol., 16,59 (I 952).


Fisher, Ronald A., "The Design of Experiments"

(T.ondo11, 1947).

' Ii-eyries, John Maynard, ",4 Treatise on Pro7?xbility" (T,ondon.'1921 ).

Tenn, John, "The

Logic of Chance"

(London, 1866).



-+ .A.














X, theoretical peak for iqitm-helical P - P vector






is described in detail elsewhere4. The Patterson DEoXYRIBONUCLEATE function is reproduced in Fig. 1. The theoretical curve for the Patterson function of a smooth helix

The very

height c = 4 on t'he fibre axis






of radius 9.0 A. is also shown (curve (i) ).

st,rong peak at


Wheatstone Physics Laboratory, King's London, W.C.2

are suggests separated. that the by phosphate approximately groups 4c. of the We two therefore chalins

added to Fig. 1 curve (ii), which completes the

st,ructure for sodium deoxyribonucleate con- theoret,ical Patterson function for two smooth

by a ~0-aXialhelices separated by &C (C is t'he pitch of the

diad a.xis. VTe have shown2 that the main features of helix). Curves (i) and (ii) t'ogether clearly Pass

W ATSON AND CR,ICKI have proposed a

sisting of two c~-a.~ialhelical chains related

t,heir st,ructure are consistent, with certain important

tlwough a large proportion of the im~ortanbPatt,erson

but P~~~~P~~.~'~groups equally spaced a~lOngsuch helices, the a,greemenhis even better.

The only n@ar-meridiOnalreflexion

in the

dia~amof st'ructure A is a rat'her weak One the eleventh layer-line. This suggests that there are


Further, we may SuPPose that t'he

Moreover, if we take into account the fact

features of our X-ray dia,gra.ms of structure B (t,he peaks.

high-humidity less-ordered folm of the salt,). A subsequent closer investigation of density and water content in relation to the prominent equatorial spacing, and also of equatorial intensities calculated from a pr~ject~ionof t,he proposed structure (kindly provided by Wat,son and Crick), makes it clear that

in detail the struct,llre is not, corlsistent with the eleven nucle~t~idesper turn of the helix. (This is

and sufficiently near to the number ten, found for structure

B1495,for the reversible transformation A + B to be

compact helical structure in which the phosphorus

that the real structure contains, not smooth helices,


equatoria,l reflexions.



considerations lead



favour a


atoms lie on a helix of radius about 8.5 A. rather

than 10 A.

This value &Iso lies within the range

sprea,d of the more diffuse layer-line peaks.

resolved peak at 5-7 A. from the pseudo-origin peak

of at c = represents a P - P vector, and this distance is exact'ly t,hat between neighbouring phosphorus

, we are more concerned bere, however, with atoms if eleven are equally spaced along one turn of

evidence which confirms in principle the type of structure suggested by mTatson and crick, than with

criticism on points

a helix of radius 9 A. In Fig. 1 we have marked wit'ha cross the theoretical ~osit~ionsof intra-molec~lar

of detail. Patterson peaks to be expect'ed for this asrangement,

assuming that t'he phosphorus atoms on one helix vertically above those on the other. The agreement

obtained pith the observed Patt'erson peaks is seen to be

good. In Fig. 2 we have marked on the Patterson function

ifsa 2-chain helical molecule exists in



exist irr st,ructure A

such a molecule, in modified form, mtlst also

(the crystalline form

at 75 per cent relative humidity), since the change A + B is readily reversible. ~h~ purpose of t,his


communication is to -point to evidence for a 2-chain helical molecule in structure A. obtained when, in the unit cell described below, there





P - P

we have measured the positions


cylindrically averaged Patterson function3. The work

For struct,ure A

and intjensitiesof reflexions and have calculated

*Sow at nirkbeck


rington Square, London, W.C.I.

Cryst~llographicI,ahoratory, 21


is one molecule (t'hat is, two chains) associated with each latt'ice point'. The circles in Fig. 2 have each 0111~

the weight

the crosses in Fig.


1/44 of the weight of t,he pseudo-lat'tice points

(re1at)ingphospnate residues only) shown by crosses


NO. 4369

July 25, 1953


15 7

near the half-cell height in Fig. 2. The lattice points relrtting the whole structure are denoted by larger crosses near to the origin level in c. In addition to this weighting, for correlation with the cylindrically averaged Patterson function, the weight of each calculated peak should be considered as inversely proportional to x, since in, averaging it must be spread over a circle of radius,~.When these factors are taken into account, it will be seen that agreement between calculated and observed Patterson peaks is still good.

Using the

Space group, density and the unit cell.

&n (in A.)

cylindrical Patterson function, we have been able to identify the lattice vectors (indicated in Fig. 2) and Fin. 3 (i) hence to index all 66 observed reflexions. 'TI-re unit

cell is face-centred monoclinic

kzving a = 22.0 A. ;

b = 39.8 A.; c = 28.1 A.; P = 96B0. If the unit cell is truly monoclinic, the space-group can only be C2, since the molec~~lecontains asymmetric carbon atoms. Again, since the structure of the phosphate sugar backbone chain is non-centric, the symmetry axis cannot pass through such a chain, but must relate the chains one to another in pairs. There must, therefore, be an even number of chains associated with each lattice point. In order to calculate the number of nucleotides in the unit cell, it is necessary to know both tlie density and the water content. Unfortunately, these quan- tities can onlv be measured on a nolycrvstalline mass, there must remain stme uncertainty as to the

true density and water content of the crystal~ites. Our measuiements5 gave a density of 1.4'7 gm./c.c. at 75 per cent relative humidity and a water content of about 40 per cent of the dry weight (or about 8H,O per nucleotide). We have confirmed that the specimens used for these measurements were in the form of structure A and not structure B. These results correspond to twenty-three nucleotides per lattice point, which is in reasonable a.greement with the suggestion that the primitive unit cell contains eleven nucleotides on each of two chains.


Fig. 3 (ii)


86 (in d.)

Three-dimensional Patterson function of crystalline soiium

a - b plane at (i) c = 0, (ii) c = 4


Sections in

was calculated. This will be published and discussed

a- b sections at heights

c = 0 and c = 4 are shown in Fig. 3. The very strong peak at a = b = 0, c = suggests a, pseudo- halving of the unit cell. On the other hand, all features of the zero section other than the origin peak are almost entirely reversed in the sectiori c = 3. This at once suggests that only a part of the structure repeats at c = Q. This is exactly what would occur for two co-axial chains related by a diad axis, as

suggested by Watson and Crick. The phosphate

groups repeat

since the two chains run in opposite directions, this

will not be true of the rest ofthe molecule.

in detail elsewhere. The

at c = 5, as indica-ted in Fig. 1 ; but,


establishing the unit cell and indexing the reflexions,

the complete three-dimensional Patterson function



Patterson function.

Relationship between 8tructure A


structure B.

In conclusion, we suggest that the unit in structure A is, as in structure B, two co-axial helical chains r~mningin opposite directions. In the change from B to A the n~unberof residues per turn increases from ten to eleven and the pitch of the helix decreases from 34A. to 28 A. In structure A the phosphate groups lie on a helix of radius 9 A., with a separation of approximately 4c between the phosphate groups of the two co-axial chains. The vertical component of the inter-base distance is thus decreased from 3.4 A. in structure B to 2-56 A. in structure A. This indicates that in structure A the planes of the bases cannot be perpendicular to the fibre axis. The positions of the strong reflexions on the sixth, seventh and eighth layer-lines suggest that the angle of tilt is about 25'. We are grateful to Prof. J. T. Randall for his interest and encouragement. One of us (R. E. F.) acknowledges the award of a Turner and Newall


[June 25.








Fig. 2. Cylindrical Patterson function of crystalline sodium deoxyribonucleate. 0, theoretical peak for iibter-helical P - P vector

Watson, J. D., and Crick, F. H. C., Sature, 171, 737 (1933).

%Franklin,R. E., and Gosling, R. G., xature,

MacGillayry, C. H., and Bruins, E. &I.,defa Cryst., 1, 156 (1918). 4 Franklin, K. E., and Gosling, R. G., Aetn Cfyst. (in the press, Part 11). 5 Franklin, R.E., and Gosling, K. G., dctcl Cryst. (in the press, Part I).

171, 710 (1953).