REPORT ON THE STOMATOPODA. 5 



frustrates the attempt to trace the gradual growth and metamorphosis of the larvae in 

 this way, and there is no other resource except comparison. 



The first step in this direction is to trace the history of each larval tjrpc, by the 

 selection and comparison of those larva which belong to the same series. In accomplish- 

 ing this I have been guided in part by general resemblances, but more especially by 

 comparative measurements. After I had tabulated the measurements in millimetres of a 

 number of specimens, which resembled each other quite closely, and formed a tolerably 

 complete series, I failed at first to trace through the columns of the table any such 

 conformity to a general law as I had expected, but more careful examination indicated 

 that this might be due to the fact that the history of the larva consists of metamorphosis 

 as well as growth, and that the size of one organ might, when compared with that of 

 another organ, show a gradual decrease during the successive stages, while its absolute 

 size was actually increasing. I therefore reduced all my measurements to a common 

 standard, and expressed them in thousandth parts of the total length of the larva at 

 each stage instead of in millimetres, and I found that this at once introduced order 

 where all had before been confused, and that, when thus reduced, the measurements 

 usually enabled me to decide with confidence whether a given larva does or does not 

 belong to a certain series. 



In a few cases these comparative measurements gave proofs of specific identity which 

 could hardly be made more conclusive by rearing the larvae. Thus the lengths of the 

 series of Coronis larvae shown in PI. XIII. figs. 1-8 are as follows, and if the length of the 

 first stage be successively multiplied by five-fourths of itself, and this number by five- 

 fourths of itself again, and so on, we obtain the series of numbers given in the second line, 

 and as it is not conceivable that an accidental collection of larvae should exhibit such 

 exact conformity to a numerical law, we may feel certain that these larvae are genetically 

 related, that they belong to one species or else to closely related species, and that the 

 scries is consecutive, with the exception of one missing stage before the last. 



After one or two series had been traced out in this way, the general character of the 

 metamorphosis itself became a trustworthy guide for establishing the series for a closely 

 related species, and thus simplified the labour, and the next step was the reference of 

 each larval type to its proper adult genus. 



If the difi"erences between the lar%'ae are due to secondary modification, we should not 

 expect the larvae of two distinct adults to become modified in the same way, and 

 although it is of course possible that the larvae of two closely related adults might 



