May 19, 1904] 



NA TURE 



65 



tto put aside the first 100 papers and to begin marking 

 ithe fresh papers on such different lines as would, in his 

 judgment, produce an approximation to the normal curve. 

 On the supposition that he had achieved that result for 

 the second 100, and continued to find that his curve was 

 pretty constant for the third 100, fourth 100, and so on, he 

 was asked at the end of all the papers to re-mark the 

 first 100. 



You might imagine that many examiners disliked having 

 to place themselves upon this bed of Procrustes, but in the 

 generality of cases it was not so. They positively took a 

 ■delight in e.xamining themselves. The process became one 

 of self-education in marking. 



Before leaving this part of my subject I should like to 

 warn you that certain causes, which an examiner cannot 

 always control, may make it difficult to obtain such an 

 ideal curve as I have shown. It is not possible for me to 

 enter fully into this part of the subject, but I will point 

 out one cause at least that he can control — I mean the 

 •examination paper. 



Good niarliing will not compensate for a bad paper. 

 Every candidate must have his chance, in some question or 

 ■other. Otherwise the examination is like a hurdle-race in 

 which the hurdles are so high that a considerable number 

 ■of candidates find themselves stopped from reaching the 

 ^oal at all. The curve, in such a case, tends to assume 

 ■a shape of this kind, mounting very rapidly to the zero 

 line (Fig. 5) — just the 

 curve, in fact, which we 

 have already seen in con- 

 nection with a dictation 

 paper. In this case it is 

 not the marking which is 

 wrong, but the examin- 

 ation paper. 



Accordingly, I found in 

 practice that it was neces- 

 sary to point out to ex- 

 aminers, before ever their 

 papers were proposed in 

 manuscript, that they 

 ought to divide their ques- 

 tions roughly into (say) 

 three portions, of which 

 one portion could be 

 answered by candidates of 

 inferior power, a second 

 should be within the range 

 of mediocre candidates, and 

 a third only possible to 

 candidates who might be 

 classed as good to e.xcel- 

 lent. The result of these 

 directions was that e.x~ 

 .aminers soon found little difficulty in spreading out their 

 •candidates in the desired way. In setting their questions 

 ithey had before their eyes the little gendarme's hat. 



Among the causes, beyond the control of the examiner, 

 ■which may interfere with the formation of this curve, we 

 must reckon as in the first rank : — (i) such a small number 

 •of candidates as does not give fair play to the law of prob- 

 abilities I (2) any selection of candidates by a preliminary 

 •examination or other means. 



With regard to the causes just named, I will only say 

 •that it has been found that the method can be applied 

 successfully when there are not less than one hundred 

 candidates, and that, even below this number, the curve, 

 though irregular in formation, gives us very useful inform- 

 .ation as to an examiner's capacities. With regard to the 

 second cause, a great deal can be done to produce a satis- 

 ■factory curve by setting such questions in the further papers 

 as are only addressed to the candidates who remain after 

 the preliminary sifting. 



I trust that I have now fulfilled the promise with which 

 I started, namely, to show you how examiners themselves 

 may be examined ; and not only this, but you will under- 

 stand that it is possible to educate examiners so as to 

 •enable them to form a much more accurate and sustained 

 judgment of a large number of candidates than would have 

 teen within their power without such preliminary 

 guidance. 



NO. 1803, VOL 70] 



Number of marks gained. 

 Fig. 5. —Dictation Paper. 



THE CHEMICAL REGLLATION OE THE 

 SECRETOR ) ■ PROCESS} 

 T^HE researches which we wish to bring briefly before the 

 Royal Society deal with the mechanism of adaptation 

 to changes in the food and the chemical correlation of the 

 activities of different organs engaged in the digestion and 

 assimilation of the food. 



According to Pawlow, the secretion of the pancreatic 

 juice is exactly comparable to the secretion of saliva, and 

 is effected by a nervous reflex. The starting point of this 

 refle.K is the stimulation of the duodenal mucous membrane 

 by the chyme, or by substances such as oil, ether, or oil 

 of mustard. Not only is the pancreatic juice turned out 

 into the intestine just at the time when it is required, but, 

 according to Pawlow, the composition of the juice varies 

 according to the food, the proteolytic ferment being in- 

 creased by a diet of meat, while the amylolytic ferment is 

 increased by a starchy diet. This adaptation of the 

 glandular activity was_ ascribed by him to a species of 

 " taste " in the mucous membrane. It was imagined that 

 the different constituents of the food excited different nerve 

 endings, which, in their turn, caused reflex activity of 

 different mechanisms in the pancreas itself. The field of 

 these assumed reflexes was considerably narrowed by the 

 researches of Popielski {Gazette Clinique de Botkin, 1900) 

 and Wertheimer {Journal de Physiologie, vol. iii. p. 335, 

 1901), who showed that the introduction of acid into the 

 duodenum was productive of secretion even after destruction 

 of all nerve connections of the pancreas and alimentary 

 canal with the central nervous system, and even after ex- 

 tirpation of the sympathetic ganglia of the solar plexus. It 

 was with a view to determine the mechanisms of this reflex 

 secretion of the pancreas, as well as of the adaptation of the 

 pancreatic secretion to variations in the food of the animal, 

 that we began our researches. 



The last named authors had also shown that the secretion 

 occurred, but in smaller quantities, if the acid was inserted 

 in any part of the small intestine, with the exception of 

 the lower end of the ileum. It was thus easy to examine 

 the effects of the introduction of acid into a loop of ileum 

 in which all nerve connections with the pancreas, or with 

 the rest of the body, had been destroyed. This crucial 

 experiment had, curiously, not been performed by previous 

 workers in the subject. On carrying it out, we found that 

 destruction of all nerve connections made no difference to 

 the result of introducing the acid. The pancreatic secretion 

 occurred as in a normal animal. It was therefore evident 

 that we had to do here with a chemical rather than a nervous 

 mechanism. Previous work had narrowed the question 

 down to such a degree that the further steps were obvious. 

 We knew already that the introduction of acid into the 

 blood-stream had no influence on the pancreas ; hence the 

 acid introduced into the intestine must be changed in its 

 passage to the blood-vessels through the epithelial cells, or 

 must produce in these cells some substance which, on access 

 to the blood stream, evoked in the pancreas a secretion. 

 This was found to be the case. On rubbing up the mucous 

 membrane with acid, and injecting the mixture into the 

 blood-stream, a copious secretion of pancreatic juice was 

 produced. It was then found that the active substance, 

 which we call secretin, was produced by the action of acid 

 from a precursor in the mucous membrane, probably in the 

 epithelial cells themselves. Once formed by the action of 

 acid, it could be boiled, neutralised, or made alkaline, with- 

 out undergoing destruction. The precursor of the substance 

 {prosecretin) cannot be extracted bv anv means that we 

 have tried from the mucous membrane. Even after coagu- 

 lation of the mucous membrane by heat or alcohol, how- 

 ever, secretin can still be extracted from the coagulated 

 mass by the action of warm dilute acid. 



We have not yet succeeded in determining the chemical 

 nature of secretin, though we have obtained chemical 

 evidence which will serve to exclude certain classes of sub- 

 stances. Thus the fact that it will stand boiling shows 

 that it is neither a coagulable proteid nor a ferment. It 

 is soluble in 90 per cent, alcohol in the presence of ether, 

 but it is insoluble in absolute alcohol and ether. It is 

 slightly diffusible through animal membranes. It can be 



1 Abstract of the Croonian Lecture. By Dr. W. M. Bayliss, F.R.S.,and 

 Prof. E. H. Starling, F.R.S. Read before the Royal Society, March 24. 



