5o6 



NATURE 



[September 22, i. 



numbered glass bottles, each bottle being provided with a con- 

 stant supply of clean sea-water by means of a system of glass 

 syphons. Into each of these bottles I placed a crab from the 

 beach. After a considerable number of deaths had occurred, a 

 series of crabs was finally established, each crab living in a num- 

 bered bottle, until it had cast its shell. The process of moulting 

 involves no distortion of the carapace, which could affect the 

 measurements concerned, and therefore each cast shell was care- 

 fully measured. The measurements of these shells were carefully 

 compared with measurements of wild crabs of the same size, and 

 the mean frontal breadth of these shells was a little less than the 

 mean breadth in wild crabs of corresponding length.^ 



After each crab had moulted, it was left in its bottle until it 

 had grown and had hardened a new shell. It was then killed 

 and measured, and the measurements obtained were compared 

 with measurements of wild crabs of corresponding size. This 

 time the captive crabs were unmistakably broader than wild 

 crabs of their own size, and there were a few of the protected 

 crabs which were very remarkably broad. The distribution of 

 abnormalities before and after moulting is shown in Fig. 6. 



This is precisely the result which we ought to have obtained, 

 if the hypothesis suggested by the study of mud were true. By 

 protecting crabs through a period of their growth, we ought to 

 raise the mean frontal breadth, and to obtain a greater per- 

 centage of abnormally broad crabs, and that is what we have 

 seen to occur. 



Of course, this experiment by itself is open to many objec- 

 tions. The estimate of age by size is a dangerous proceeding, 

 and it is difficult to exclude the possibility that confinement in 

 a bottle may directly modify a crab during the critical period of 



It would take too long to go into that matter now, and I shall 

 not attempt to do so. I will only now ask you to consider oiie 

 or two conclusions which seem to me to follow from what I have 

 said. 



I hope I have convinced you that the law of chance enables 

 one to express easily and simply the frequency of variations 

 among animals ; and I hope I have convinced you that the 

 action of natural selection upon such fortuitious variations can 

 be experimentally measured, at least in the only case in which 

 any one has attempted to measure it, I hope I have convinced 

 you that the process of evolution is sometimes so rapid that it 

 can be observed in the space of a very few years. 



I would urge upon you in conclusion the necessity of ex- 

 tending as widely as possible this kind of numerical study. 

 The whole difficulty of the theory of Natural Selection is a 

 quantitative difficulty. It is the difficulty of believing that in 

 any given case a small deviation from the mean character will 

 be sufficiently useful or sufficiently harmful to matter. That is 

 a difficulty which can only be got rid of by determining in a 

 number of cases how much a given variation does matter ; and 

 I hope I have shown you that such determination is possible, 

 and if it be possible, it is our duty to make it. 



We ought to know numerically, in a large number of cases, 

 how much variation is occurring now in animals ; we ought to 

 know numerically how much effect that variation has upon the 

 death-rate ; and we ought to know numerically how much of 

 such variation is inherited from generation to generation. The 

 labours of Mr. Galton and of Prof. Pearson have given us the 

 means of obtaining this knowledge : and I would urge upon you 

 the necessity of obtaining it. For numerical knowledge of this 

 kind is the only ultimate test of the 

 theory of Natural Selection, or of any 

 other theory of any natural process 

 whatever. 



Fig. 6.— Distribution of abnormaliiy of frontaj breadth ratios_ in 527 female crabs before and after 

 moulting in captivity. The contirucus line shows the distribution before, the dotted line after 

 moulting. 



moulting, and so on. All these points would have to be dis- 

 cussed at greater length than your patience would bear, before 

 we could accept this experiment by itself as a proof that some 

 selective agent exists on the shore, which is absent from the 

 bottles. At the same time, the result of this experiment is 

 exactly what we should expect to find if such a selective 

 agent did exist, and so it is in complete harmony with the 

 evidence already put before you. 



Of course, if the observed change in frontal breadths is really 

 the result of selection, we ought to try to show the process by 

 which this selection is effected. 



This process seems to be largely associated with the way in 

 which crabs filter the water entering their gill-chambers. The 

 gills of a crab which has died during an experiment with china 

 clay are covered with fine white mud, which is not found 

 in the gills of the survivors. In at least 90 per cent, of the 

 cases, this difference is very striking ; and the same difference 

 is found between the dead and the survivors in experiments 

 with mud. 



I think it can be shown that a narrow frontal breadth renders 

 one part of the process of filtration of water more efficient 

 than it is in crabs of greater frontal breadth. 



1 This was probably due to the death-rate during acclimatisation beirg 

 selective. It was very difficult to keep the apparatu$ clean ; and the 

 deaths which occurred were in most cases due to the presence of putrescent 

 bits of food, which had not been removed. 



A subsequent experiment was made with the fame apparatus, in which 

 crabs were kept in putrid water until a large perctntage bad dieo : and the 

 mean frontal breadth of the survivors was found to be distinctly less than { 

 the mean frontal breadth of the dead. ' 



SECTION G, 

 mechanical science. 

 Opening Address by Sir John 

 Wolfe Barry, K,C.B., LL,D., 

 F.R.S., President of the 

 Section. 

 Apart from all the other considera- 

 tions which so favourably affect this 

 Congress, I think, so far as Section G 

 is concerned, that we are fortunate in 

 meeting in this ancient city, which has 

 so much of special interest for en- 

 gineers and for others interested in 

 applied science, 

 (i) I propose, therefore, to say a few introductory words 

 about Bristol and its neighbourhood from the point of view of 

 this section of the Association, but it is far from my intention 

 to either criticise the past work of the Corporation in relation to 

 their dock enterprises or to volunteer advice to them with re- 

 spect to possible works of improvement. 



Bristol is, at this moment, of great commercial importance, as 

 indicated by the value of its imports and exports, and occupied 

 an even more important relative position among British ports at 

 a time when the ports of Liverpool, Glasgow, Cardiff, or South- 

 ampton were almost, or altogether undeveloped. So far as 

 Customs Revenue is concerned Bristol now stands third, and 

 in regard to the gross value of her sea-borne trade she is 

 thirteenth among ports of the United Kingdom. 



It is unnecessary, and it would be foreign to the objects of 

 Section G, to attempt to trace the economic reasons which have 

 caused the long- continued importance of Bristol, or to account 

 for the rapid growth of other ports more or less competitive with 

 her. All such causes are to be found, at least to a great extent, 

 in considerations apart from the merely physical characteristics 

 of the sea, river, or land at the various sites, as, for example, in 

 propinquity to markets or centres of production, in situation 

 relatively to population or to means of distribution, in individual 

 or collective enterprise, in enlightened or unenlightened 

 administration. 



These circumstances have^ in truth, at least as much if not 

 more influence in determining the history and prosperity of 

 ports than what are termed natural advantages of respective 

 sites, by which I mean such matters as protection frcm winds 



NO. 1 508, VOL. 58] 



