13.2 report— 1865. 



origination, but the diffusion — of ideas. Books and newspapers and reviews, no 

 doubt, are the main agents for doing that work. Still it is, I think, indisputable 

 that, as seeing is proverbially more impressive than hearing, so what we hear 

 orally delivered makes upon us a stronger impression than that which lies on a 

 printed page on which our attention may or may not dwell. The other is the 

 stimulus given to inquiry by the mere fact of investigations of this kind, or the 

 result of them, being brought prominently and conspicuously before the public. 

 Men go home with their heads full of subjects on which they perhaps never thought 

 seriously before ; and since, as I believe, nothing once known is ever really forgotten 

 — since an idea which has once found lodgment in the mind, though its presence 

 there may long have been barren, and though we ourselves may have been uncon- 

 scious of it, will often spring up into life after a long interval — it is difficult to 

 determine what crop will not grow, sooner or later, out of the seed thus cast about 

 apparently at random. 



And now let me say a word as to the right application of the statistical method. 

 To use figures rightly, assuming that they are accurate in themselves, is not so 

 easy a matter as it is apt to appear. There are various fallacies into which 

 unpractised statisticians fall, one or two of which may be worth noting. One, perhaps 

 the commonest, arises from the use of too narrow a basis for calculation. To express 

 my meaning, — given a certain class of men, between certain ages, as soldiers, agri- 

 cultural labourers, artisans, and the like, the aggregate length of life among 10,000, 

 or even among 1000 of them, will be practically a fixed quantity, you can determine 

 it beforehand. But the length of any single life is of course uncertain ; and if so 

 few as 10, 20, or even 100 lives be taken, the element of chance, or what we call 

 chance, is not sufficiently excluded, and a single exceptional case affects the general 

 result. Another error, less easy to detect, arises from not taking into account all 

 causes which affect the result. Thus, supposing that the question is, the effect 

 produced by reduction of a tax in increasing consumption of the article taxed, it 

 is natural, but a very obvious mistake, to urge, "post hoc, ergo propter hoc," to speak 

 of that augmented consumption as arising solely from the reduction of duty which 

 preceded it, ignoring the causes — such as general prosperity, and consequent aug- 

 mentation of the consuming power ; cheapening of means of transport, or lowering 

 of the actual cost of production by mechanical improvements, or otherwise of the 

 article consumed. Or, to take an illustration from a different class of subjects : — 

 Suppose it is desired to ascertain the average mortality of a certain class ; that for 

 this purpose we take the mortality in that class during a certain number of years, 

 and that, during one of those years, an epidemic of a destructive character has pre- 

 vailed, you of course strike out that year from the series as unduly affecting the 

 average. But, if you stop there, your calculation is again vitiated, though in a 

 different manner, for it will almost always happen that an epidemic sweeps away a 

 larger proportion of unhealthy than of healthy lives, and so the years succeeding their 

 mortality, having been, so to speak, anticipated, are above the true average in poii^t 

 of health, and do not form a firm basis for a permanent calculation. Another variety 

 of the same error, identical in principle, is when effect and cause are confounded. 

 For instance, I have heard a sanatorist argue, with great general truth, in favour 

 of model-lodgings in towns, and support his argument by pointing out the 

 diminished rate of mortality in them, as compared with that of the same class out- 

 side. No one could dispute the general conclusion, but in fairness it should have been 

 taken into consideration, as a qualifying proposition, that model-lodgings, by their 

 reputation for superior cleanliness and healthiness, attract not the average, but the 

 best men of the class for whom they are designed — that is, those whose chances of 

 life are the best, independent of the room in which they are lodged. Hence in- 

 voluntary exaggeration, by which a good cause could only suffer. 



Another and perhaps simpler example is the existence of a hospital, or sanato- 

 rium, in an otherwise healthy locality. Unless that be allowed for, it is clear that 

 the place in question will be credited with many deaths which are not due to it, — 

 nay, this absurdity follows, that the more the spot is resorted to by invalids (Madeira 

 would be a case in point) on account of its very healthiness, the higher its apparent 

 mortality becomes. This last example is not without its practical importance, 

 especially in cases of military mortality. It is obvious that, by a system of dis- 



