Sf.ptk.mueii 14, 1883.] 



SCIENCE. 



375 



But coming down to a much later period, we find a 

 rcniiirkablo application of the law of induction in a 

 worlv upon the industries of France, by the minister 

 Chaptal. He presents airricuhural tables, which have 

 been received with great coiili<leiice, since they bear 

 tlio appearance of oflicial statistics, and were exe- 

 cuted under tlic Empire. Ilis tables arc found to 

 have been computed, without acl<nowledgnient, from 

 a statement addressed by M. Hennet, director of the 

 cadastral survey, in 1817, at a time when not more 

 than a seventh part of this work had been finished. 

 The other six-sevenths were obtained by a simple 

 multiplication of the finished part. 



Many years ago, a ' distinguished statistician ' 

 published, with great apparent precision, tlie yield of 

 potatoes in France. There had been no official in- 

 ventory taken; but when one came to be made, some 

 time afterwards, it was found that this deduction 

 had been obtained by niutiplying the yield of a single 

 commune by 37,000, the number of communes in 

 France. 



These examples might be multiplied indefinitely; 

 and we need not cross the Atlantic, nor go far back in 

 time, to find them. There is scarcely a day, but that 

 we see passing through the newspapers, estimates, 

 deductions, and statements, that have no more solid 

 foundation than tliose that we have cited. Never- 

 thcloss, we must not wholly disregard the inductive 

 method in statistics: there are many eases in which 

 we can get nothing else. 



The chemist must analyze the soils and the ores 

 from samples. In every operation of testing the qual- 

 ity and the value of any commodity whatever, we 

 must select from the material before us what appears 

 to be the average quality. And so of statistics gen- 

 erally: if there is no actual and general inventory 

 made, we must collect from what is deemed a fair 

 average, .and, from these data, obtain such conclusions 

 as they afford. The result in this, as in every thing, 

 will depend upon the intelligetice and honesty of the 

 person who makes the estimate, the extent of his 

 opportunities, his experience, and his skill. 



Keturning to the field of exact statistics, we may 

 remark, that we can never have an accurate census 

 of the population until we have a thorough and tmi- 

 forni registration of births, marriages, and deaths; 

 a measiu'e which this association undertook to pro- 

 mote, more than a quarter of a century ago, but 

 which has not made successful progress. 



We cannot have a faithful statement of the indus- 

 tries, without a record kept of the production, the 

 consumption, and the cost of operation. This is 

 already done by most of the important ones, as an 

 incident of business; but we lose the advantages by 

 the hurried manner in which the oflicial inquiries are 

 made. Yet upon these returns we rely for all that is 

 collectively known about them. 



It follows, that, until we can realize these desirable 

 features, the best we can expect is, to afford more 

 time for previous preparation, by submitting before- 

 hand tlie questions that are to be answered; which 

 can only be done by the aid of ' householders' sched- 

 ules' for population, on 'special blanks' for each of 



the iiulustries, or other subjects, that'come within the 

 range of inquiry. 



It was my intention to dwell at some length upon 

 the illustration of statistical facts by graphic methods: 

 but time will not permit, and opportunity for full 

 preparation has not been found. For more than 

 thirty years I have bi^en accustomed to note down the 

 principles involved in these methods, whenever, in 

 the course of a wide and varied range of oi)portunity, 

 a new one was found; and it has been with nie a 

 cherished intention to present the whole subject in 

 a systematic form. 



We may concisely state, that graphic illustrations, 

 using lines, areas, or angular spaces, often supple- 

 mented by colors, may be employed for representing 

 either — 



1. Quantities, with or without reference to time. 



2. Time, in recurring, interrupted, or progressive 

 periods. 



3. Direction, or relative position ; aiul 



4. Intensity or force. 



In general, but two elements can be clearly pre- 

 sented at once; but by a skilful use of different 

 colors, or kinds of lines, subjects of the same nature 

 may be admirably comi)ared, and the relations of 

 cause and effect not only illustrated, but even discov- 

 ered and proved. It is often admissible to introduce 

 subjects having di-similar notation. — as, for exam- 

 ple, degrees of temperature, and height of barometer, 

 — in the same drawing; but in these cases each must 

 have its own scale, and, in a general way, every dia- 

 gram must have a scale for every element of the sub- 

 ject that is represented, either expressed or implied. 



Quantities may be shown either as they exist at 

 certain periods of time, or as they form parts of a gen- 

 eral total ; and they may be presented so as to exhibit 

 successive subdivisions, down to any desirable degree. 

 If the divisions of a general total do not require sub- 

 division, they may best be shown by angular spaces, 

 as sectors, which together make up the whole of a 

 circular area. If the divisions have some qualities in 

 common, the shades of color may be of different in- 

 tensity, significant of the degrees of quality that may 

 exist. But where there are successive subdivisions, 

 or parts of parts of a whole, there is no way so exact 

 as by means of rectangular areas, which may also be 

 shaded in different tints, as well to separate them one 

 from another as to show differences of intensity or 

 degree. 



in both of these methods, as well of angular spaces 

 as of rectangular areas, we can only show quantities 

 .-IS they exist at a given point of time. We catch, as 

 it were, the coiulilions, as does the light, the image in 

 a camera. They admit of no such thing as motion 

 or change; but these changes may often be strikingly 

 rei>resented by a series of diagrams, presenting the 

 conditions at different periods of time. 



Where time and qu.anlity are combined, we have 

 an easy and striking means of illustration ; and in this 

 the time may be in recurring periods, such as the 

 hours of a day, or the months of a year, or it may be 

 progressive, as in a series of years. 



For the recurring periods, I think there is nothing 



