August 5, 1922] 



NA TURE 



179 



is seen in the case of the pea-crab and various bivalve 

 molluscs and ascidians, or Myzostoma and Antedon. 

 Now many jelly-fishes are infested with amphipods 

 such as Hyperia, and it is a reasonable deduction 

 that these crustaceans may be feeding on the food 

 material collected by the jelly-fish. Thus, if the young 

 fishes which take shelter below the umbrella of a jelly- 

 fish assist the jelly-fish by keeping down ecto-parasites 

 such as Hyperia, then an intelligible explanation is 

 offered of the association of young fishes with such an 

 apparently voracious host as a large jelly-fish, for in 

 return the jelly-fish in these circumstances would have 

 less of its own food stolen. J. H. Orton. 



Oyster Store, Packing Shed Island, 

 West Mersea, July 2, 

 and The Laboratory, Plymouth, July 14. 



Roche's Limits for Satellites. 



The notice in Nature (July 15, p. 89) respecting Dr. 

 Fountain's work on Roche's limit for satellites brings 

 to my recollection some estimates which I made many 

 years ago with respect to the stability of satellites 

 moving close to the surface of Mars (Trans. Roy. 

 Dub. Soc, 1897, v °l- vl ■)■ 



The question arose in connexion with a theorv 

 accounting for the " canals " as resulting from stresses 

 set up in the surface rocks of Mars by the proximity 

 of such satellites. The doubling of the canals came 

 out nicely and the curvature of the canals as mapped 

 by Lowell, Douglass, and Pickering was in agreement. 

 But the doubt arose as to whether former satellites 

 of sufficient magnitude could have preserved their 

 stability when circulating around the planet with 

 the requisite degree of approximation. 



Assuming that the satellite possessed the cohesive 

 strength of basalt and taking the case of Phobos 

 supposed to be moving in an orbit but 23 miles from 

 the planet's surface (i.e. with but five miles separating 

 the surfaces of planet and satellite), I found that the 

 satellite, even at this distance, would be stressed only 

 to one-seventh of its breaking strength. 



J- Joly. 



Trinity College, Dublin, Julv 15. 



Optical Definition and Resolving Power. 



In Mr. Mallock's letter on " Definition, Resolving 

 Power and Accuracy," published in Nature of 

 May 27 (vol. 109, p. 678), reference is made to the 

 measurement of star images on eclipse plates, from 

 which one might infer that the evidence for the 

 Einstein deflexion of light obtained in 1919 was of a 

 very doubtful character. 



I have had no experience in measuring star images ; 

 but there is little doubt that if the same order of 

 accuracy can be obtained as is possible in measuring 

 spectrum lines, the Einstein deflexion should be 

 easily determinable with a focal length of 19 feet, 

 provided that it can be disentangled satisfactorily 

 from the scale correction. 



In my method of measuring photographs of spectra 

 the image of a line is not bisected by a thread, but 

 a positive copy is superposed on the negative in the 

 micrometer and the coincidence of the two images 

 estimated. By this means the intervals to be 

 measured are doubled, and an extraordinary degree 

 of precision is attainable with practice, as is shown 

 by the agreement between different measurers. I 

 have often had occasion to repeat measures made by 

 one of my assistants of the shifts of the solar lines with 

 reference to the arc lines, and we rarely differ by an 

 amount exceeding o-ooi mm. in the interval measured. 

 This is the result of taking the means of six settings in 



NO. 2753, VOL. I io] 



each line, and the probable error derived from the 

 accordance of settings is usually about half a micron. 



Probably star images cannot be measured so 

 accurately as this by the ordinary method of bisection, 

 but a skilled measurer should be able to determine 

 the position of a star easily within 0-005 mm. or, 

 on the scale of the eclipse plates, within o" 18. 



J. Evershed. 



Kodaikanal, June 24. 



Interspecific Sterility. 



Dr. Bateson in his letter on interspecific sterility 

 (Nature, July 15, p. 76) seems to lay insufficient 

 emphasis on certain facts. If one considers plant 

 and animal species in general, it would appear that 

 interspecific sterility is by no means so general as 

 was formerly assumed to be the case. Among the 

 OEnotheras, in which great numbers of species 

 crosses have been made, complete fertility, in the 

 sense that large numbers of fertile offspring are 

 produced, is the rule unless the forms differ in 

 chromosome number. Even species of Oenothera 

 which come from widely separated regions and differ 

 conspicuously in all their characters, including 

 flower-size, are fertile in crosses. That a certain 

 amount of gametic and zygotic sterility also frequently 

 occurs is of course well known, and it is probably 

 correctly interpreted in terms of lethal factors. But 

 lethal factors are not peculiar to wild species, for 

 numbers of them arise in the mutations of Drosophila 

 melanogaster. 



Among animals, interspecific sterility appears to 

 be more widespread, but even here the Bovidae are, 

 I believe, all interfertile. The contrasted condition 

 of the Equidae, at least as regards the horse and the 

 ass, is accounted for by the difference in their 

 chromosome numbers. In the Drosophilidae, where 

 interspecific sterility is extreme, there is a considerable 

 range in chromosome form and number. The two 

 species, Drosophila melanogaster and D. simulans, 

 which are extremely alike and have similar chromo- 

 some groups yet produce sterile hybrids, might be 

 cited as corresponding exactly to Dr. Bateson's 

 conception of interspecific sterility. But it is an 

 extreme case, and there are probably more numerous 

 instances to cite on the other side. 



Dr. Bateson refers to the case of Oenothera gigas 

 and agrees that tetraploids frequently do not breed 

 freely with diploids. But he says that " the applica- 

 bility of that example is exceedingly doubtful " 

 because we "can scarcely regard an unresolved pair 

 of twins, such as the tetraploid must be, as a 

 specifically distinct organism." It is this statement 

 in particular to which I should be inclined for 

 several reasons to take exception. In the first place, 

 in calling the tetraploid an " unresolved pair of 

 twins " Dr. Bateson scarcely recognises the intimate 

 character of the union involved. I formerly analysed 

 (Arch. f. Zellforsch. vol. 3, pp. 525-552, 1909) the 

 changes which have occurred in CE. gigas in so far 

 as this could be done by comparative cell measure- 

 ments, and found that the cell units were not merely 

 larger, owing to the doubling in the chromosome 

 content of their nuclei, but that in various tissues 

 they were altered in shape, the increase in one 

 dimension having been much greater than in another. 

 Moreover, the genetic behaviour of CE. gigas indicates, 

 as de Vries first contended, that some other change 

 has taken place in the germplasm of this species, in 

 addition to the doubling of the chromosomes. 



I have only recently been convinced on this point 

 by comparisons of CE. gigas with the tetraploid 

 forms obtained by Winkler (Zeits. f. Botanik, 8, 



