374 



SCIENCE. 



[Vol. 11. , No. 32. 



material for a considerable number of deductions, — 

 as, for example, the population sheet of the census, — 

 it is generally best to take but one tliiiu/ at a time. 

 Thus, the ages, professions, nativities, civil condition, 

 etc., should be taken by separate operations, and not 

 two or more at once. There is not, however, the least 

 need of confusion in keeping the subdivisions of these 

 subjects, in two or four classes, — as, for example, 

 ages by sex and color, — by a simple arrangement of 

 heavy and light horizontal lines upon the tally-sheet, 

 and a little practice in its use. 



Thei'e is much to be gained, both in time and accu- 

 racy, by a proper arrangement of a tally-sheet. The 

 grouping together of tally-marks, by making four 

 down and one across, so as to divide the work into 

 groups of fives, is so natural and obvious a method, 

 that few who have had occasion for this kind of work 

 could have failed to adopt it. By an arrangement 

 which I have used to a large extent in census work, 

 I have had my tally-sheets printed off into squares, 

 so that each compartment should receive one group 

 of five, and no more. Then, by a series of numbers 

 with a common difference of five, printed across the 

 top of the sheet, at the head of each vertical column, 

 the number of tally-marks in a horizontal row can 

 be known at once, by glancing along the vertical col- 

 vimn containing the last full group of fives, to the 

 number printed at the top, and then adding the marks 

 in excess, but less than five, in the next compartment. 

 This saves all counting, and a considerable amount 

 of time. There is also an advantage, on account of 

 the eye-sight, in having the tally-sheets of some other 

 color than white: a neutral tint might be best, but I 

 have found common manilla paper answer every pur- 

 pose. 



Plans have been proposed for using cards of differ- 

 ent sizes and colors, properly inscribed or numbered, 

 as counters, for classifying a variety of facts, forming 

 together a definite whole. By using colors, the eye 

 becomes, without mental effort, a guide to the hand, 

 in their distribution into piles or cells in a case; and, 

 when the work is done, their number may be accu- 

 rately known by weighing, or by measuring the height 

 of each pile. Those of different sizes could be sepa- 

 rated by mechanical devices, without handling, and, 

 by a little practice, without liability to error. 



It may be said generally, that the chief, indeed the 

 only real, difficulty, in the preparation of statistical 

 data, consists in getting the facts correctly. There is 

 nothing in the operation of a central ofiice that needs 

 to involve eri'or; or, if an error is committed, there 

 should be no difficulty in tracing it to the clerk who 

 is responsible for it. An efiicient way to secure accu- 

 racy in work would be, to make a money-charge 

 against the clerk who commits an error, to be paid to 

 the one who finds it. I believe that something of 

 tlie kind is done in some of the statistical offices in 

 Europe, a class of revisers being employed, who are 

 paid by the fines thus imposed upon the careless. 



With respect to statistics obtained by circulars ad- 

 dressed to persons supposed to have the information 

 desii'ed, we have every grade of value, from good to 

 good for nothing. The result depends upon many cir- 



cumstances: as, for example, whether the person 

 making the return is paid, or is under some obliga- 

 tion to, or expects some favor from, the person or 

 office making the inquiry; whether the inquiry can 

 be answered by reference to a record, or by some 

 research more or less conveniently made, or is to 

 be supplied from personal opinion, and a general 

 knowledge of the subject; or, finally, wlielher the 

 question can be answered by any thing better than a 

 guess, by one who knows perhaps very little about it. 



I would hold it to be the general rule, that where 

 the inquiries are few and simple, exact as to their 

 object, and, if they refer to a record, exact as to time 

 and subject, and especially if they can be returned 

 upon the same blank, and without expense for postage 

 or otherwise, a very large percentage will be answered 

 without a second application. A repeated call would 

 probably hi'ing a third or a half of the remainder; 

 but there will be, now and then, one who will fail 

 to reply, unless under official or personal obligatiou 

 to do so. 



We have thus far considered the dealing with sta- 

 tistics that have been gathered from the whole of a 

 given field of inquiry: there are other methods that 

 deserve notice, and the first of these is that ' by 

 samples.' A portion of some whole is carefully 

 studied, and the results obtained are deemed ajipli- 

 cable to the entire field. 



The French statistician Moreau de Jonnes has 

 given some instances of this method, as applied in 

 times past, by persons who had acquired eminence, 

 and whose work gained confidence ; and very prop- 

 erly asks, ' What is such work worth ? ' 



Vauban, distinguished as a military engineer, at 

 the beginning of the eighteenth century, wishing to 

 know the agricultural production of France, and the 

 revenue it would yield, resorted to a method which 

 would appear strange enough now, but still may be 

 called ingenious. lie attempted to i-each his object 

 by taking an exact account of the production of a 

 squai'e league, reckoning the arable land, vineyards, 

 pastures, and woodlands, with their products in 

 quantity and value; and then, by the simple "rule of 

 three," he said, " as 1 is to 25,000, so is the result to 

 the whole of France." 



The English agriculturist, Arthur Young, sought 

 to ascertain the proportions of meadow-land, mount- 

 ains, and the like, in France, by cutting up a map by 

 lines following these features of the surface, and 

 weighing the parts. 



In 1700 Lavoisier, distinguished in science, and 

 f(5r this reason consulted by the national asseinbly 

 upon a question of imposts, found no existing data 

 that applied to the internal resources of the counti'y, 

 until he himself supplied them, by a method that is 

 now altogether neglected 'in statistical researches. 

 He proposed to ascertain the number of ploughs in the 

 country, and from this to calculate the quantities, 

 production, and consumption of agricultural crops. 



In 1784 M. Necker, the distinguished statesman, 

 deduced the population of France from an assumed 

 percentage of the birth-rate; and this was taken for a 

 census! 



