Skptember I/, 1896] 



NA TURE 



461 



of much utility to science. If the height of the barometer be 

 correlated with the frequency of explosions in mines, it would 

 not appear utile to describe the barometer as a " cause " of the 

 exjilosion. Or, to take another case, which is purely hypo- 

 thetical, but which will, I think, illustrate Prof. Ray Lankester's 

 point. There is, we will suppose, a purely random distribution 

 of supernumerary teats in cows. But in my particular herd the 

 two best milkers possess supernumerary teats (although there is 

 no correlation between such teats and good milkers in general). 

 I keep the calves of these two cows because they are good 

 milkers, and by reason of this selection supernumerary teats 

 become more and more common in my herd. At last I begin 

 to preserve calves with supernumerary teats, really because this 

 is a test of their descent from the good milkers, practically 

 because I find them in themselves good milkers. Now, because 

 I am a careful breeder, my cows may get a reputation at sales 

 all over the country, and a correlation between supernumerary 

 teats and good milkers may come to be generally recognised. 

 This happens not because supernumerary teats are a cause of 

 good milkers, but owing in the original instance to a random 

 association of this variation with a utile variation. Thus, a 

 primarily random association with a favourable variation may 

 liy the principle of heredity quite easily lead to a correlation 

 which it would be of no profit to consider causal. If two 

 characters be not correlated, and one be favourable to survival, 

 then any selection of the favourable character, which hits a 

 group of individuals having more than the average of the second 

 character — and this may easily arise if we breed from com- 

 paratively (ew individuals — will by the principle of heredity 

 lead to & fortuitous correlation. I do not assert that this is the 

 case in the frontal ratio of crabs, but it seems to me that a link 

 is really missing in the chain of demonstration. All causality 

 is of course correlation, but the converse, which Prof. Weldon 

 seems to hold and Prof. Lankester to controvert, is surely a 

 dangerous doctrine ? Karl Pe.arson. 



September 10. 



Specific Characters among the Mutillidse. 



Thk discussion in your columns as to the utility of specific 

 characters leads me to offer a few remarks on the Mutillidie, an 

 interesting family of Hymenoptera. In the arid region of the 

 United States, this family is very numerously represented, and 

 its members may be seen running about in warm weather, 

 especially frequenting sandy places, roads and pathways. It is 

 not at first apparent why the species should be so numerous, 

 living under what seem to be identical or almost identical 

 conditions; in 1893 {Trans. Aimr. Ent. Soc, xx. 343), I 

 wrote: "It is difficult to account for the origin of so many 

 species under conditions which can hardly at any time have 

 been very diverse." But the region in question is inhabited by 

 very many species of bees, the modifications of which have 

 relation to a varied flora, as I have illustrated by particular 

 instances elsewhere (Proc. Ac. Nat. Sci. Phila., 1896, pp. 

 33-41). The various Mutillidx are parasitic in the nests of 

 these bees, and consequently do not live under identical con- 

 ditions : we have a varied flora with its varied insect-visitors, 

 and these with varied parasites, the whole series of phenomena 

 intimately connected, though at first sight it would seem 

 impossible to see any connection between the flowers and the 

 nuilillids, however indirect. 



It must be a long time before the actual host-relations of all 

 the mutillids are known, but I have a species now under 

 ol)servation, which may serve as an example. The bee 

 Diadasia diininuta lives in colonies, burrowing perpendicular 

 tunnels in the beaten pathway, which are produced somewhat 

 above the level of the ground by ineans of fragile cylinders of 

 .sandy particles, designed to keep the tunnels from being filled 

 with sand. The little Splucrophthahna heterochroa is the 

 parasite of this bee, and the females may be seen in numbers 

 running about between the burrows, now and then looking into 

 them or entering. At once we see the utility of one of the 

 specific characters ol S. heterochroa — its small size. The larger 

 species could not enter the small burrows of the bee. 



The females of S. heterochroa are splendid little insects, 

 ornamented with scarlet, black, and whitish. Like the females 

 of all Mutillida>, they are wingless. The much more active 

 winged males, which are not so elegantly ornamented as the 

 females, may be seen bustling about, looking for the latter. In 



the Mutillidae, the females are very varied in colour, markings 

 and structure ; while the males are much more uniform. Thus, 

 Cameron says ("Biol. Cent. Amer. Hymenoptera," p. 259): 

 "This general re.semblance of the males not only makes their 

 specific determination a work of difficulty, but it adds greatly 

 to the task of assigning them to their respective partners of the 

 opposite sex." If the bright and varied colours and markings 

 were due to activity or a " katabolic tendency," it is in the 

 winged males that they ought to be found ; not, as is actually 

 the case, in the wingless females. But on the principle of 

 utility there may be an explanation. The males have to look 

 for their respective females, and I believe the ornamentation of 

 the latter assists their recognition. 



There is a whole series of Mutillidos which are very plainly 

 coloured, from tawny through various shades to black, never 

 with any scarlet, or conspicuous markings. These {Photopsis 

 and Brachycistis) are all nocturnal, without any exception, and 

 come to lights in the evening. But the systematists who have 

 described many of these insects, were totally unaware of this 

 circumstance until I pointed it out recently ! 



The moral of all this is, that to understand the real meaning 

 of specific characters we must study the species in nature. We 

 are hardly more likely to understand natural phenomena from 

 the examination of dead animals alone, than a Hottentot would 

 be to understand the apparatus of telegraphy. And eventually, 

 I believe even the pure systematist will have to base his work oiv 

 biological observation. It has been fondly hoped all along that 

 absolute criteria of specific distinction would be found in the 

 insects themselves, without reference to their habits ; and the 

 searchers for such " hall-marks," driven from point to point, 

 have at length taken refuge in the male genitalia. But only a 

 few days ago I received the following in a letter from M. Ernest 

 Andre, the distinguished French student of Hymenoptera, and 

 particularly MutillidEe. 



" Comme je i'ai dit, je crois qu'on attache aujourd'hui une 

 trop grande importance aux caracteres tires de I'appareil genital 

 male. Ces organes sont tres variables, difficiles a apprecier, et 

 ne concordent pas toujouts avec les autres caracteres morpho- 

 logiques." T. D. A. COCKERELL. 



N.M. Biological Station, Mesilla, New Mexico, 

 U.S.A., August 25. 



The Khmer of Kamboja. 



In Nature, June 11, p. 135, I see a short notice of the 

 work being done in Australia by Prof. R. Semon, of Jena, and 

 that he classes the "Khmer and Chams of Kamboja" as 

 "primitive Dravido-Australians." I hope that some of your 

 anthropological experts will, as soon as possible, correct this 

 serious mistake. 



Mr. A. H. Keane, in Nature, January 6, 1881, p. 222, 

 calls these people " Caucasian " ; but I presume they are 

 now (1896) known to be what Captain C. J. F. S. Forbes 

 classed them, i.e. Mon-Anam, in his " Languages of Further 

 India." 



J. R. Logan, in his " Ethnology of the Indo-Pacific Islands," 

 published at Singapur and Pinang, in the /oiirnal of the Indiai> 

 Archipelago (1847-63), rightly classes them as Mon-Anam, 

 giving their linguistic peculiarities and alliances, pronouns, &c.,, 

 and numerals up to ten, those from one to five being identical 

 with our Kol, Sontal, Munda, Ho (the most western relatives 

 of the Mon-Anam alliance), and quite different to the Dravidian, 

 numerals. 



The now civilised Kambojans admit that the " Khmer dom " 

 are the older and purer stock, whence they are descended, and 

 that they were hill savages, which carries out what we know- 

 so far of these early /«-Burmo-Tibetan races from the Asam 

 side. 



Both in physique and languages, the Dravido-Austral 

 aboriginal of India (south of Himalaya) and the Mon-Anam 

 group are very distinct ; the former are seen purest in the 

 Andamani and Negrito, in whom there is an entire absence o£ 

 Tibetan elements. 



But the Mon-Anam (which includes the Khmer) were formed, 

 as a race, by the mixture, or fusion, of (Sifan) Tibetans with 

 the Dravidians, lying south of Himalaya from Nipal to East 

 Asam. M one time this " Mon-Anam" race appears to have 

 covered all Northern and Eastern Bengal, the whole of Asam, 



N(J. 1403, VOL. 54] 



