St. Joseph, Missouri, claimed, one year, a rate of 6.5 deaths out of every 1,000 inhabitants. Were this figure authentic, the thriving Missouri city, by the law of probability, should be full of centenarians. It isn't. I essayed to study the local reports, hoping to discover some explanation of the phenomenon, but was politely and regretfully informed that St. Joseph's health authorities issued no annual reports. The natural explanation of the impossibly low rate is that the city is juggling its returns. In the first place, that favorite method of securing a low per capita death rate--estimating a population greatly in advance of its actual numbers--is indicated; since the community has fewer lines of sewers and a smaller area of parks than other cities of the size it claims--two elements which, by the way, would in themselves tend to militate against a low mortality. Perhaps, too, the city has that ingenious way of eliminating one disturbing feature, the deaths under one week or ten days, by regarding them as "still-births." Chicago used to have this habit; also the trick of counting out non-residents, who were so thoughtless as to die in the city. At present, it is counting honestly, I believe. Buffalo used to pad for publication purposes. One year it vaunted itself as the healthiest large city in the country. The boast was made on the original assumption of a population nearly 25,000 in excess of the United States Census figures, to which 20,000 more was added arbitrarily, the given reason being a "general belief" that the city had grown to that extent. Perhaps as complete returns as any are obtained in Maryland, where the health official, Dr. Price, culls the death notices from 60 papers, checks up the returns from the official registrars, and if any are missing, demands an explanation by mail. It behooves the registrar to present a good excuse. Otherwise he is haled to court and fined. The Board has thus far never failed to secure a conviction. Now, if the most concrete and easily ascertainable fact in public health statistics, the total of deaths, is often qualified or perverted, it follows that dependent data, such as the assigned causes of death, as required by law, are still more unreliable; so I shall keep as far away from statistics as possible except where some specific condition can be shown by approved figures or by figures so inherently self-disproved that they carry their own refutation. _The Criminal Negligence of Physicians_