222 



SCIENCE 



[N. S. Vol. XXXVII. No. 945 



are carefully arranged, are well written and ap- 

 parently have been tried out with elementary 

 classes. These chapters form the most satis- 

 factory part of the work. 



In chapter XXII., Classification of Insects, 

 keys are given by which the student is sup- 

 posed to be able to determine insects to orders 

 and families. Here the authors have not been 

 so fortunate. 



The classification of the Hymenoptera 

 adopted (p. 319) is woefully inadequate even 

 for a very elementary text, and does not repre- 

 sent any of the progress made within the last 

 quarter of a century. Granting the diversity 

 of opinion that still prevails in regard to 

 many points of the classification of this order, 

 there is, however, much that has been done 

 within that time that can not properly be 

 ignored. 



The inclusion of saw-flies, gall-flies and 

 parasitic Hymenoptera in one suborder, as 

 contrasted with all the remaining groups 

 (aculeates) in another, is indefensible. The 

 inclusion of Lydidse and Xyelidae in the Ten- 

 thredinidEe, and of Cephidse, Xiphydriidse and 

 Oryssidse in with Siricidse has no longer any 

 justification. On the other hand, the best 

 founded work on ants to-day recognizes but a 

 single family, comprised of five subfamilies. 

 It would seem that this position might at 

 least be advantageously maintained in an ele- 

 mentary key which, for the sake of brevity, 

 omits mention of many important families. 

 Neither this course was followed nor the alter- 

 native of including five family groups. On 

 the contrary, three family names appear in the 

 key, Poneridffi, Myrmicidfe and Camponotidae. 

 The Dorylidse, relatively unimportant in this 

 country, as well as the very important Doli- 

 choderidse, are not mentioned. The latter in- 

 cludes the economically important Argentine 

 ant (referred to in the text on page 259 as 

 belonging to the Myrmicidse) as well as sev- 

 eral very common North American forms 

 which the casual student is far more likely to 

 frequently meet with than he is with any 

 Poneridse. By the key all the Dolichoderidas 

 would fall into the so-called Camponotidse, 

 with which they have no closer relation than 



have any of the other groups of ants. The 

 term Camponotidm for the group containing 

 the genus Formica is used for Formicidas, 

 despite the fact that the superfamily name 

 Formicina based on that genus is used directly 

 above it — a rather astonishing neglect of the 

 established customs as well as codes of nom- 

 enclature. 



There may be some excuse for omitting 

 from the key such families as Evaniidse, 

 Stephanidse, Trigonalids, Sapygidas and Mel- 

 linidffi, but to be consistent the equally un- 

 common Masaridse should be omitted. But 

 why is there no place provided for the Lar- 

 ridae, Nyssonidse, Philanthidse, Pemphredon- 

 idse and Crabronidse, all of which are abun- 

 dant in species and individuals? The ma- 

 jority of the species belonging to these fam- 

 ilies fall, according to the key, in the Bem- 

 becidas, the others in Sphecidse, but the divi- 

 sion is along a line that can make no pretense 

 of being natural or even convenient. 



It might also be noted that there is no pro- 

 vision made for wingless Hymenoptera in the 

 keys and that therefore the wingless parasitic 

 species, the worker ants and female Mutillidse 

 can not be classified. The term Psammo- 

 charidae is used for Pompilidse or Ceropalidae 

 of older authors, but the name Proctotrypidse 

 is not replaced by Serphidse. 



The whole principle of key construction 

 with the intention of omitting " less impor- 

 tant " forms is open to grave objection. The 

 result is that the student chancing upon a 

 specimen of a non-included group (and in the 

 present case such specimens will be legion) 

 ends in running it out to a family to which 

 it does not belong, and confident in the cor- 

 rectness of his labors and unsuspecting the 

 untrustworthiness of the key, has it impressed 

 upon his memory as one of the types of that 

 family. A far better method, in fact the only 

 defensible method where keys are presented, 

 is to make them complete enough to provide 

 correctly for all forms coming within their 

 scope, and then if deemed desirable rare or 

 less important groups may be bracketed or set 

 in special type. Otherwise it were better the 

 tables were omitted altogether. 



