256 MESSRS. C. SHEARER, W. DE MORGAN, AND H. M. FUCHS 



could be reared through metamorphosis. This led to the attempt to discover definite 

 characters in tlie larval development, more specific and fixed than those hitherto 

 employed as a basis of comparison between normal and hybrid plutei. Moreover, 

 of the three species selected for our experiments, two, E. esculentvs and E. acutus, 

 are closely related ; while the third, E. miliaris, is but distantly allied to the other 

 two, being usually classed in the separate genus of Psammechinufi. Thus, if the 

 hybrids between E. iniliaris and E. esculentus, and E. miliaris and E. acuhis should 

 prove sterile, on account of the specific difierences separating these species, those 

 at least between E. esculentus and E. acutus would, we hoped, yield fertile hybrids, 

 which could be investigated in a second generation. 



In the numerous papers which have appeared on Echinoderm hybridization, hitherto 

 only the characters of the early pluteus, up to the assumption of the eight-armed 

 condition, have been investigated. In the past it has been held that the early larval 

 characters are sufliciently definite to give a clear answer to all questions of inheritance. 

 The larval skeleton has been adopted as a main basis of compai'ison between normal 

 and hybrid larvae ; unfortunately, the skeleton, like many of the early larval features, 

 is subject to irregular variation. Under unfavourable circumstances the skeleton 

 frequently tends to develop extra rods and sjsinous processes, which give it a spurious 

 resemblance to that of other species. Therefore skeletal characters cannot always 

 be safely attributed to hereditary influence. In our experiments we soon came to 

 the conclusion that if any definite advance were to be made in this subject, it would 

 be best to abandon the skeleton, and all the early features of the pluteus, as an index 

 of parental influence, and to find more definite characters in the later larval life, and 

 that we should have rigorously to adhere in our work to two general principles. 



First, the characters which we should investigate in the hybrids must be perfectly 

 definite and subject to as little variation as possible in the parent species ; any wide 

 variation of these characters in the pure-bred species would render their study in the 

 hybrid forms of little value. If this variation in the pure-bred species overlapped, 

 then no clear results could be obtained. The most definite conclusions can only be 

 drawn from the investigation of characters which are invariably present in one parent 

 while absent in the other. 



Secondly, the experiments should be conducted in sucli a manner as to render it 

 reasonably certain that all the conditions surrounding the growth and development 

 of the hybrids should be as far as possible normal. It is only too clear that, in the 

 past, investigators have not observed this condition, and too often have been at 

 fault in drawing conclusions from experiments that gave obviously stunted and 

 deformed larvae. In such cases the absence of a character is frequently due to 

 a pathological condition. It is essential, therefore, that the technique of rearing 

 the hybrids should be such as to obviate as far as possible all unnatural conditions. 



The first of these requirements demands a thorough knowledge of the normal 

 embryology of the forms to be investigated, while the second requires the knowledge 



