avoided. Physicians have taken considerable interest in censuses, and for the very reasons stated, and so in many cases health districts have been prescribed and the statistics of population and the social facts relating to population for such health districts preserved. In this way the very best results are to be reached. With complete statistics of population for clearly defined health districts, where the sanitary conditions can be compared and differences of conditions noted, a scientific study of death-rates with reference to the density of population can be undertaken. The ordinary statistics of death-rates based on the density of population of cities are exceedingly vicious, but perhaps not more so than the ordinary statements relative to the death-rate of cities based on the whole population. There is great liability to very misleading statistics in this direction. The errors arise from two causes. The first of these is the incompleteness of death statistics. This can only be overcome by a compulsory registration of deaths. The second cause is that population is not accurately known except for periods some distance apart, and here error arises, and would arise, even with complete and perfect statistics of deaths; as, for instance, a State which depends entirely upon the Federal census ascertains its population only once in ten years. For the census year the death-rate based on population may be fairly accurate; but for intermediate years the death-rate must be based upon calculations of population mathematically made. In some cases this has led to very vicious results, and has caused considerable fright and anxiety on account of the great apparent death-rate, when, had the facts all been known, it would have been found that the death-rate was really normal. Another feature of error, or rather feature for the basis of erroneous conclusions, relative to the death-rate in great cities, arises from the fact of the existence of large hospitals in cities, and that the death-rate is increased by people coming from the country to the cities for treatment and there passing away thus giving an abnormally high death-rate relative to the actual living population of a city. This is also true in connection with the criminal statistics of cities. Men come in from country towns for the purpose of a visit or a spree, or for carrying out some nefarious design. At all events, they commit crime, from one cause or another, within the city limits, are there arrested and punished and their crimes help to swell abnormally the legitimate criminal statistics of the city itself. All these considerations should be taken into account when writers are undertaking to draw what they feel to be accurate conclusions through comparisons of statistics. I have read very learned essays upon conditions of the population, involving insanity, crime, disease, death, etc., when all the conclusions of the essays were based upon most incomplete
