chusetts. But, from time to time, improvements have been made. The most important change was made in May, 1885, since which time the representative physicians in active general practice who report regularly each week, are asked to report only the sickness which occurs. under their own observation. The system is thus on a more accurate scientific basis than when the reports included sickness which occurred in the vicinity, under the observation of other physicians; because the opportunities for conferring with other physicians vary greatly, and thus cause variations in the reports which are eliminated by confining the reports to answers to direct questions, relative to each important disease, did you or did you not, during the given week, see a case of that disease? How many cases did you see? This system of sickness statistics rests upon the "law of averages," in accordance with the "law of probabilities," which makes it probable that a considerable number of representative physicians in active general practice, in localities distributed fairly well about a State, will see an average of the sickness which occurs. Theoretically, this was true before the system was inaugurated. Practically, it has been demonstrated to be true, by the several combinations of the sickness statistics which have been made in Michigan. For instance, it was questioned whether the small number of reports received in time for the weekly bulletin, being about fifty, were sufficient to supply a fair statement of the sickness in Michigan, throughout the State. The question seems to have been answered in the affirmative as follows: When the fifty reports are compiled, and afterwards fifty more reports are received and compiled, it is generally found that the results of the two compilations are practically the same; indicating that both results are correct. Again, speaking now of the whole number of reports received and compiled in the Annual Report, although during the years 1877-84 the system was perfect as since May, 1885, yet such diagrams as that on page 255 of the Annual Report of the Michigan State Board of Health for the year 1886 (reproduced further in this article), prove that the results of the compilation of the weekly reports of sickness then yielded statistics of very great value, and statistics which bore on their face the evidences of reliability, being consistent from month to month, and from year to year. Not only that, but when placed beside the results of tri-daily observations of such "instruments of precision" as the standard thermometers in use by the regular meteorological observers in Michigan, these sickness statistics relating to the prominent diseases bore an astonishingly close resemblance to the statistics of the observations of the thermometer. The similarity of the two curves, representing the two sets of observations-of the sickness and of the instruments, is such as to lead to the conviction that both sets of observations are true, and that there is a necessary relation between the sickness and the atmospheric temperature. Until the sickness statistics in Michigan sup not as plied the scientific basis for such studies, such knowledge as that to which I have just referred was not in existence, as a scientific induction; it rested upon the insecure foundation of dogmatic assertions, not supported, as it now is, by large numbers of facts classified and tabulated Again, diagrams have been constructed containing curves representing the sickness reported from several important diseases, and these curves have been compared with curves representing the mortality from those diseases during the same time in Michigan. These comparisons have demonstrated that both kinds of statistics are valuable; although it showed, as was known before, that the mortality statistics in Michigan are not as complete during the earlier as they are during the later months in each year, for reasons which have been published, and have long been well known to those who understand the vital statistics of Michigan.* It is to be hoped that the legislature will be induced to give this subject attention, and will so amend the law for the collection of the mortality statistics as to require that each death shall be immediately recorded, as the sickness now is, within a week of its occurrence, and not be as the deaths now are, neglected for a year or so before being placed on record. It is apparent, to whoever will take the trouble to investigate the subject, that the sickness statistics of Michigan are extremely valuable. The vast mine of valuable information has been only very slightly worked as yet, for the reason that few private individuals can find the time to do such careful painstaking work for such long-continued researches as are required to so marshal the vast members of facts as to bear upon a question in the causation of disease. Such work for the benefit of all should be done once for all by some one representing the whole people. Enough has been done by the Secretary of the State Board of Health to prove that the sickness statistics of Michigan are a mine of wealth which will richly repay the work of developing. The part which has been developed is that relating to the diseases of the throat, lungs and air-passages, and those dangerous communicable diseases which enter the body by way of the throat and air-passages. It has been demonstrated that the causes which increase these diseases are associated with coldness of the atmosphere, and that the relation is not only qualitative. This information alone is of sufficient value to repay all and more than has been expended for the statistics. But the "cold-weather diseases " are only one part of all-there remain to be studied the great classes of diseases which prevail most under opposite conditions. There remain also the studies into the exact conditions which are causative of each disease, and those which are only accidental. And, after all the factors shall have been worked out, there will remain the study of the rate of decline of each disease under the influence of the appropriate meas* Explained on pages 3-12, and 158-167. Vital Statistics of Michigan, 1871. ures which shall be inaugurated for the prevention or restriction of each disease. Consequently, it should be a long time before the collection, compilation, and utilization of sickness statistics should cease. Should such work ever cease? It would seem not; because even if the causation of all the prominent diseases shall become known, the preventive and restrictive measures thoroughly applied, and the diminution of all preventable diseases shall reach its maximum, there should still be constant surveillance of such sickness as shall remain, in order to be prepared to enter at once upon measures which may at any time be indicated as necessary. There is another reason than the one already stated why we may not expect the sickness statistics to be immediately used by people generally, but only by statisticians or persons who are willing to work hard and thus grasp the principles of statistical methods. That reason is found in the fact that the sickness statistics do not impress the average person, especially not the casual observer, as being sufficiently reliable upon which to base questions involving human life on a grand scale as is demanded by public sanitation. To some of the physicians who contribute the weekly reports, it seems almost or quite incredible that from the experience of one person like himself in a place, and comparatively few places in Michigan, can be built up statistics which shall enable one to know the exact relations of all the most important and common diseases to the varying conditions of the atmosphere. Perhaps as useful a service as can be done, at the present time, is to point out how it is that the fifteen years experience with the sickness statistics of Michigan has resulted in the development of statistics which are not only the best that the world has ever seen, but that can be demonstrated to be perfectly reliable.* Some such demonstration here follows: It is extremely difficult for any person to master the subject of vital statistics. This should not be wondered at, because all questions relating to the life of man--the most complex of all organized beings-are complex; and the laws or principles of even ordinary statistics are not generally understood. To hear statistics spoken of as unreliable, is a very common occurrence; but it is true, and ought to be generally known, that no general fact is established or can be established except by the statistical method. One experience or one observation of a fact does not establish anything; a few experiences or observations may lead to a belief; but it is only by the grouping together of oft-repeated ex *The confident tone in some portions of this report may be accounted for by the fact that the member of this committee was, for a few years, engaged in compiling and studying general statistics, and for twenty-one years has been continuously engaged in compiling and studying several kinds of vital statistics. Accordingly, it is assumed that this long experience and study of statistical science and practice, especially of vital statistics, enables him to test and judge of the reliability of such statistics. periences or observations of facts that a general fact is established; and the grouping together of experiences and observations of facts, is the essential part of statistics. Therefore there is no higher method, no more reliable method, than the statistical method. That method is the one employed in every science. For instance, the chemist makes an experiment, and observes a result. In order to reach a conclusion, however, he must repeatedly observe the same result to follow repetitions of the same experiment. If, because of the imperfections of all manipulations, he does not always reach precisely the same results, he continues, by the statistical method, to learn the amount of the probable error, and, if that is within the limits of variation warranted by the nature of the manipulation, he uses the average result, and if the observations are sufficiently numerous, he relies upon that result; that is to say, if he has (what is really, although he may not so consider it) a statistical basis for his conclusion he is satisfied. Not only the science of chemistry, but all sciences rest upon similar foundations. One of the most common fallacies in the minds of intelligent people, is, that statistics are not reliable unless all the facts on the given subject are collected. Concerning vital statistics, this false idea is extremely common, even among scientific men who in their own science never seem to think of the comparative paucity of the facts upon which their own conclusions are based. A chemist, for instance, will base conclusions on less than a hundred experiments, sometimes less than a dozen, yet, in vital statistics, he would be likely to question results of observations of many times that number of experiences; and very likely demand that records be used of all the possible experiences. Few people realize how perfect is the knowledge which may be gained by statistics which embrace only a small fraction of the great mass of facts which might be collected on the same subject. Take, for instance, in the science of meteorology, the statistics of temperature; it is well known that thermometers hung on the different sides of any building will rarely exactly agree. But it is not so well known that if one intelligent and reliable observer take tri-daily observations of a thermometer placed under proper conditions, in a central location, as, for instance, in Lansing, the results of his observations supply a reliable basis for conclusions concerning the entire State of Michigan, as to the comparative temperature in each month of every year. This is susceptible of demonstration, so that any intelligent person can appreciate the result; and a demonstration for a single year is given on page 27 of the Annual Report of the Michigan State Board of Health for the year 1886, and a similar demonstration is in nearly every annual report. Page 27 of the report for 1886 is here reproduced, diagram 1, on page 461. By that diagram it may be seen that a curve, accurately drawn to scale, representing the exact average daily temperature at Lansing (the line —ox) is very nearly the same as the curve representing the average for twenty |