742 REPORT — 1894. 



work there are half a dozen ways of solving any particular prohleni. In some 

 fashion or other the enp;ineer must he ahle to judge between these various methods, 

 methods which are often very much alike, but each of which may possess certain 

 particular advantages and certain particular drawbacks. The arithmetical criticism 

 which merelj' counts the advantages and the drawbacks, and puts an equal number 

 of the one against an equal number of the other, is common enough, hut obviously 

 useless. The ■^ery first necessity to the critic is that he should have what I have 

 just called the sense of proportion, a sense which will enable him to distinguish 

 mere academical objections from serious practical difficulties, which shall euable 

 him to balance twenty advantages which can be enumerated on paper by one 

 serious drawback Avhich will exist in fact, which will enable him in fact to place 

 molehills of experience against moimtains of talk. It is perhaps a doubtful point 

 how far this sense of proportion can be taught at all. No doubt it can only be built 

 up upon some natural basis. I am sure tliat in engineering we all know men whose 

 judgment as to whether it was advisable to take a particular course we would 

 accept implicitly, because we know that it is based on large general criticism, in spite 

 of the most elaborate and specious arguments against it set down on paper. Any 

 third-year student — not to go still further back — can criticise perfectly along 

 certain very narrow lines, just as anyone can learn the rules of harmony and can 

 write something in accordance with them which purports to be music. But after 

 all the music may be music only in name, and the criticism may not be worth the 

 paper it is written upon, however formal it may appear to be, unless the writer is 

 thoroughly imbued with a sense of the proportionate value of the different points 

 which he makes. To take the commonest possible case, I dare say we have all 

 of us heard certain methods, mechanical, chemical, or other, stigmatised as totally 

 wrong and absolutely useless because they contain certain easily provable errors. 

 I am sure, too, that most of us could give illustrations of cases in which this has 

 been said with the very greatest dogmatism when the errors of the impugned 

 method are not one-tenth part as great as the equally unavoidable errors of obser- 

 vation in the most perfect method. 



Probably the best special education in proportion which a man can have is 

 a course of quantitative experimental work. I say qitantitative with emphasis, 

 aa meaning something much more than mere qualitative work. Here, I think, 

 comes in the usefulness of the engineering laboratory. We require that the 

 training should be not only in absolute measurement, but in relative measurement, 

 the latter being quite as important as the former. Many kinds of measurements 

 stand more or less upon a level as a training of the faculties of observation in 

 themselves, but no single kind of measurement is sufficient as a training in 

 proportion. A year spent in calibrating thermometers or galvanometers might 

 make an exceedingly accurate observer in a particular line, but it would not give 

 the observer a knowledge of what even constituted accuracy in other directions ; 

 for accuracy is a relative and not an absolute term. In most engineering matters 

 the conditions are, unfortunately, of a most complex kind ; so complex that our 

 problems are incapable of anj' solution sufficiently exact to satisfy the mathe- 

 matician or physicist. The temptation to treat these problems as the mathe- 

 matician treats those with which he deals — namely, to alter the assumed 

 conditions in order to get an exact solution — is a very strong one. I am afraid 

 it is most strong often in those engineers who are the best mathematicians. It is 

 a temptation, however, steadily to be resisted. We must assume our conditions 

 to be what they actually are, and not what we should like them to be ; and if we 

 cannot obtain an exact solution of our problem with its actual conditions, so much 

 the worse for us, not so much the worse for the conditions. Our first duty is 

 generally to find out the conditions; if they are disadvantageous (in fact I mean, 

 and not merely in the problem), to alter them if they can be altered, but not to 

 ignore them because they are inconvenient. We have then to find out the extent 

 to which the known conditions permit any exactness of solution at all, and, 

 finally, we have to keep this in view as a measurement of the highest accuracy 

 which is attainable. To work out certain branches of the problem with such 

 minuteness as to give us apparently very much greater accuracy than this is not 



