Popular Science Monthly/Volume 58/April 1901/Suicide and the Weather

From Wikisource
Jump to navigation Jump to search

By Professor EDWIN G. DEXTER,


MUCH has been written and rewritten on the subject of suicide. It has long been a favorite topic with the student of social statistics, and has been scientifically treated from the standpoint of race, of nationality, of social condition, of occupation and of climate. Whole volumes have been devoted to the problem and magazine articles almost without number. It is not, however, my intention in this paper even to summarize the conclusions arrived at in all this mass of literature, but to discuss a phase of the subject which can not have escaped the reader of the daily paper, and has long proved an enigma to the special student of the problem of self-destruction—that is, the daily fluctuation in the occurrence of suicide. Why is it that upon picking up our daily paper one morning we see the heading 'Epidemic of Suicide', and find the details of six or eight or even a dozen successful or unsuccessful attempts recorded for the previous day—a number greater than for the whole week preceding? Yet such is often the case—so often, in fact, as not infrequently to have been the subject of editorial comment, with vague queries as to the cause of such a wave of emotional depression and consequent self-destruction.

The answers to this query have been many and varied, among the most frequent of which has been chance. Mimicry and suggestion have been proposed, and without doubt have their place in the solution of the problem of the periodical fluctuation of the suicide curve, but still can not account for all its peculiarities. The weather has also been suggested as the cause of the fluctuation referred to, and it is to the following out of this promising clew that this paper is confined.

From a priori grounds it would seem to be a good one, for of all the environmental conditions, those of the weather are the only ones which vary for all the individuals in a given locality simultaneously. A and B and C all have troubles peculiarly their own, the climax of which could not be expected to occur upon the same day; but when the east wind blows and the sky is leaden A, B and C all feel the influence, whatever it may be, and an empirical study of large numbers of A's and B's and C's, noting their behavior under such conditions, would seem to be the surest method of discovering just what the influence is.

That weather states have a mental effect has long been recognized. Literature is full of allusions to the fact, and not a few of the world's great thinkers have left on record their own emotional flights and depressions under different meteorological conditions. But most of us need to take no other word for the fact than our own. In all the vigor of perfect health such influence may hardly be recognized, but when the vital powers are depleted by the exhausting effects of a long nervous or physical strain, then this phase of the cosmical environment is sure to make itself felt. Then come the days when everything goes wrong. The groundwork of forgotten quarrels is remembered, uneasy questions arise with regard to the future; one gets tired of life. And how much of all this can be attributed to an east wind or a leaden sky—in other words, to weather effects? In order to answer this question we must define our use of the term 'weather effects.' From the standpoint of our present study we should include within the category of weather effects any marked inequality in the occurrence of suicide which may be found to bear a fixed relation to the fluctuations of what we call weather. We conclude that a fixed relation between a given weather state and an unusual prevalence of suicide is causal and not accidental. This is based upon an inductive study of large numbers of data, and is as valid as such studies can well be.

The problem, then, consists in discovering these fixed relations. In order to do this with exactness, the meteorologist's analysis of weather must be taken. To him a given weather state is a complex and not a simple phenomenon. He reads its temperature, its barometer, its humidity, its wind velocity, its sunshine or shade, and its precipitation, and it is only to the synthesis of these conditions that he applies the term weather. For the purpose of our present study it is not enough to say that the weather is fine, or disagreeable, or muggy, for those terms mean one thing to one person and something very different to another, so it has been necessary to make use of a definite meteorological nomenclature which is recognized the world over. The study is in no sense an attempt to account for suicide, but for the irregularity of its occurrence. Man always has sought and perhaps always will seek self-destruction as the relief for sorrow, fancied or real, and the basal reason for this is not to be found in the weather. We would not argue that the weather drives people to suicide save in very exceptional cases, but, on the strength of what follows, that under some weather states other things are peculiarly liable to drive people to the act. In other words, that some meteorological conditions so affect the mental state, so influence the emotional balance, that ordinarily endurable things become unendurable, and life seems no longer worth the living.

This problem, which seems to show a causal nexus between the weather and the mental state of the suicide, is a comparison of the occurrence of suicide under different meteorological conditions, with the normal prevalence of those conditions, noting the excess or deficiency. The data were collected for New York City and the city of Denver, Col., and although the climatic conditions of the two cities are very different, it is in no sense a comparative study for them. In fact, so few data (two hundred and sixty suicides) were procurable for the western town that but little weight is given to conclusions based upon them, compared with the much greater number for New York City, and the study of the former is only incidentally mentioned.

The method of procedure was as follows: In order to procure the proper data of suicide for the city of New York the records of the coroner for five years were carefully gone over (some 28,000 separate death certificates), disclosing the particulars of 1,962 suicides, and the exact number (varying from to 9) tabulated for each of the 1,826 days of those years. Next the police records for the same five years were studied, and the number of unsuccessful attempts for each day noted. This record is quite complete, since in the eyes of the law one attempting suicide is a criminal, and must be so branded on the books. From these two sources were obtained the exact number of persons who for each day of the period covered were of suicidal intent, unless some unsuccessful attempt escaped the surveillance of the police. In the present article neither age, sex, nationality, nor occupation is considered; simply the fact that some one wished to die by his own hand—for the five years, 2,946 in all for the city of New York.

When the data of suicide had thus been tabulated, the meteorological basis for the study was obtained from the records of the United States Weather Bureau. At the New York station (Denver for the Denver study) were copied the mean temperature, barometer and humidity, the total movement of the wind, the character of the day and the precipitation for each of the 1,826 days of the period considered, and placed opposite the already tabulated number of suicides. Then, by a somewhat laborious process of tabulation, the exact percentage of days which were recorded at the Weather Bureau under each of the seventy-seven definite meteorological conditions represented by the accompanying figures was computed. That is, the exact percentage characterized as 'clear,' as 'partly cloudy,' or 'cloudy,' as having some or no precipitation (without considering the amount), as having had a mean temperature between zero and five degrees F., between five and ten degrees, and so on for each one of the designated groups for temperature, barometer, humidity and wind. Now, it may be readily seen that these percentages represent the normal or expected occurrence of suicide for each meteorological group if the weather had no effect. For instance, if thirty per cent, of the days are found to be characterized as 'clear,' we should expect that same percentage of suicides for 'clear' days plus or minus the percentage due to probable error from accidental causes (which with the number of data used would be very small) if the character of the day had no influence on their occurrence. If forty per cent, did actually occur under such conditions, we should be forced to conclude that fair days were prolific of suicide, as indeed they seem to be. This principle was applied to each of the meteorological groups, and the figures show graphically the results.

For each, the general meteorological condition is indicated at the top; the definite group readings are given in small figures upon the heavy vertical lines which represent the occurrence of suicide for the group. Expectancy for each group is represented by the vertical distance A—B and excess or deficiency graphically shown in percentages of this, which may be read by means of the scale at the left.

The method of tabulation, by means of which the actual occurrence of suicide for each meteorological group, was determined was similar to that for expectancy, and needs no further explanation.

PSM V58 D615 Suicide and weather correlation chart.png

Fig. 1.

Monthly Distribution.—Fig. 1 indicates very wide variation in the number of suicides occurring in the different months of the year—generally speaking, the heated months showing excesses and the cold ones deficiencies when compared with the normal. May and August show the greatest numbers, with the least for February, in spite of the fact that the shortness of the last-named month is taken into consideration.

It may be seen, by an inspection of the figure, that the increase in number for each month from February to August, and the decrease for the other months of the year, would give an almost perfectly regular crescendo-diminuendo to the occurrence curve were it not for the fact that April and May are raised out of their position by unusual excesses. Why April, which in its general weather characteristics is Elysian compared with its immediate predecessor, should show one-fourth more suicides, and May, which by common acclaim is one of the most delightful of the calendar, should present a number surpassed only by sweltering August, it is not easy to see. Yet such is the case for the five years covered' by this study, and similar conditions have been demonstrated by other students of the subject. Morselli, in his exhaustive treatise for the European nations, finds that for thirty-two separate studies made by him the maximum numbers were in June eighteen times and in May eight times. In explanation of the fact he says, "Suicide is not influenced so much by the extreme heat of the advanced summer season as by the early spring and summer, which seize upon the organism not yet acclimatized and still under the influence of the cold season." There is little doubt that the end of winter brings with it a depleted condition of vitality, both nervous and physical; yet I am inclined to think that the

PSM V58 D616 Suicide and rain correlation chart.png

Fig. 2.

fact can not wholly account for the great increase in the later spring months. In the conclusion of this paper the condition is again alluded to, and at this point I would simply call attention to the fact that the increase comes with the season of the year when rejuvenating Nature is in her brightest mood.

Character of the Day and Precipitation.—The terms 'clear’ 'partly cloudy' and 'cloudy,' as used by the Weather Bureau's characterization of weather states, have a definite and technical meaning. The first is used to designate days on which the sun is obscured for three-tenths or less of the hours from sunrise to sunset; the second from fourtenths to seven-tenths of that period; and the third eight-tenths or more. (See Fig. 2.)

Under precipitation I have considered separately days which were absolutely free from rainfall or snowfall, and those on which there was either, without considering the amount.

The figure referred to discloses some unexpected facts—namely, that the clear, dry days show the greatest number of suicides, and the wet, partly cloudy days—the gloomiest of all weather—the least, and with differences too great to be attributed to accident or chance; in fact, thirty-one per cent, more on dry than on wet days, and twenty-one per cent, more on clear days than partly cloudy. As will be seen, on cloudy days the occurrence was about normal. What does this mean? Must fiction resign her right to ring in gloomy weather and blinding storms as a partial excuse for ending an existence made more unendurable by these? If such be the case, it is well that Dickens and Lytton and Poe are gone, for they would be robbed of a large number of their tragic climaxes. England has long been characterized as 'gloomy Britain,' and Montesquieu has called it the 'classic land of suicide,' stating that the 'excessive number of suicides for that country is due to its gloomy weather.' Statistics have shown, however, that the number is not excessive there, being less per million inhabitants than for any other important European nation. An interesting paper, appearing in the British magazine Once a Week (vol. xix.) over no signature (though the writer was evidently not a Scotchman), has a bearing upon the subject. It says:

"The idea that the prevalence of suicide in this country (England) is due to our bad weather is precisely one of those hasty and illogical inferences which are characteristic of the Gallic mind. The constant gloom of bad weather ought to acquaint us so thoroughly with moods of depression that suicide would never occur to us. Look at Scotland, for instance, where suicides are rare. Why are they rare? Simply because a succession of Scotch Sundays has so accustomed the people to prolonged despondency that any sudden misfortune can not sink their spirits any further. One has only to spend a dozen Sundays in Glasgow or Edinburgh to become inoculated against suicide. So far from London fogs driving people to jump off Waterloo Bridge, they ought to train the mind to bear any calamity. A man who has taught himself to eat prodigious quantities of opium feels scarcely any effect from other forms of intoxication. We can educate our mental susceptibilities as we can our muscles, and the more we educate them the more they are able to bear."

There are many truths beneath the jocular vein of this quotation, and the writer expressed more facts than perhaps he knew.

Certainly a comparison of suicides for Denver and New York City supports his theory, for in the former city, where cloudy and partly cloudy days are less than one-third as frequent as in the latter, we find suicide excessive during the gloomy weather. Yet the conditions there, both social and climatic, are so unusual as to give this fact little weight in a comprehensive study of suicides, and we must maintain that Vilemais's dictum that 'nine-tenths of the suicides occur in rainy or cloudy weather' is utterly unfounded upon fact, at least for the conditions covered by this study.

Temperature.—Fig. 3 seems. to show plainly two things: (1) That the greatest excesses of suicide are found at the two extremes of the temperature scale, when the conditions entailed the maximum of actual misery, and (2) that the next greatest excesses occur during the pleasantest conditions of temperature. I would here, however, call attention to the fact that for all the figures the readings at the extremes of the conditions are based upon fewer data than those nearer the middle, hence are more liable to accidental error. For example, although the temperature group zero to five degrees shows an excess of two hundred and

PSM V58 D618 Suicide and temperature correlation chart.png

Fig. 3.

ten per cent., the condition occurred but twice in the five years studied, and the whole number of suicides was but eight, while the excess of fifteen per cent, for the group sixty-five to seventy degrees is based upon two hundred and sixty-eight. For this reason the value of the readings at the extremes of all the figures, except Fig. 1 and the upper limit of Fig. 5, at which point there were data enough to give validity to the findings, is lessened when compared with other points in the curves.

Taking this fact into consideration, the greatest numerical excesses in suicide occur in the temperature group from forty-five to seventy degrees. This places them within the category of most agreeable temperatures, for within those limits are found the monthly means of April, May, June, September and October. The deficiencies of suicide occur in the groups from twenty to forty-five degrees, conditions which are not generally considered most agreeable and within which are found the monthly means for the colder months of the year.

These results, however, are corroborative of the findings for the study of monthly occurrence which show deficiencies for those months. The excesses for extreme conditions of heat and cold are perhaps only what might be expected. In the thickly populated tenements of the city great heat becomes so oppressive as hardly to be endured, and at the other extreme of temperature, when the mercury of the thermometer

PSM V58 D619 Suicide and barometric pressure correlation chart.png

Fig. 4.

is only in the bulb, both personal misery and a feeling of sympathy for a dependent family might prompt one to self-destruction as the last resource.

This curve does not differ materially from that of the Assault and Battery,[1] except that in the latter it is shown that for the highest temperature ever experienced those misdemeanors, as recorded by the police, show deficiencies. For them the numbers increase regularly up to a temperature of eighty-five degrees, but above that point they fall off very rapidly. This fact, however, is not hard to account for, since a considerable amount of energy is required to be objectionably out of order, and at such conditions of heat this seems hardly available.

Barometer.—Considering the liability that accidental conditions affect the validity of our curves at their extremes, the results shown in Fig. 4 prove conclusively that low conditions of pressure are accompanied by excesses in suicides, with corresponding deficiencies for the reverse barometrical readings. We can not, however, suppose that it is the actual density of the atmosphere which produces this marked effect. A difference of pressure as great as that between the two extremes for New York City would be experienced in going to the Adirondacks, and five times as great in a trip to Colorado, without producing tendencies to personal annihilation, so we must look for our explanation elsewhere. It is probably to be found in the relation which exists between atmospheric pressure and some other weather states—possibly storms. The peculiar mental and physiological conditions which prevail for a considerable period just preceding violent storms or marked changes of weather have long been recognized, and it may be that in them we have the solution. Persons afflicted with gout or rheumatism, or even corns, can 'feel' the approach of such meteorological conditions, and certain mental peculiarities are probably just as prevalent. Many weather proverbs are based upon the unusual activities of members of the animal kingdom at such times, and as a storm is often preceded by a low condition of the barometer, we have perhaps an explanation of their cause. More work, however, must be done to demonstrate this as a scientific fact.

Humidity.—The results of the study of suicide for this condition (Fig. 5) are in themselves conclusive, but directly opposite to those found in similar studies made for Assault and Battery, Deportment in the Public Schools and the New York City Penitentiary, and the behavior of the insane.[2] For suicide the excesses are for high humidities; for the others mentioned they were for low.

The showing for suicides seems to be what would be naturally expected if we were to theorize on the matter, as those unendurable 'sticky' days, when one feels it his prerogative to be 'out of sorts,' are usually of high humidity. There are some interesting conclusions to be drawn here by a comparison of this curve with that for precipitation. The latter showed deficiencies of suicide for rainy days, while this gives an excess for humid ones. Now, all rainy days are humid, but not all humid days are rainy, and our logical conclusion must be that the excesses shown by the present figure must have been for the humid variety, yet without precipitation. Such precisely is the 'sticky' weather mentioned, and its effect must have been deadly to produce such results.

In accounting for the unusual number of assaults and misdemeanors in the public schools for low humidities, as discussed in the paper cited, the electrical potential of the atmosphere for such meteorological conditions was considered the cause. It is a fact conceded by scientists that at every point upon the earth's surface there are lines of electrical force extending off into space, and that the potential is roughly in a reverse ratio to the humidity prevailing at a given time. This electrical condition for regions of universally low humidity, as the altitudes of our western plateaus, is very marked and productive of no slight effects. These

PSM V58 D621 Suicide and humidity correlation chart.png

Fig. 5.

seem to be a mental and even physical exhilaration, productive of energy which in the long-run generally proves to be in excess of the normal healthy possibilities. The result is for those regions a tendency to overwork, especially mentally, with a resulting state of collapse. Although these conditions are not so marked for the higher humidities of the sea level, they nevertheless exist to a degree, and without doubt in New York City there is less individual surplus energy when the humidity is relatively high than when relatively low. This would lead us to infer that, from the showing of this condition, suicide was excessive when energy was low. This relation of occurrence to available energy is reversed for certain of the figures, but other conditions enter in which are discussed in the conclusion of this paper.

Wind.—But little need be said upon the effect of this factor as shown by Fig. 6. The regularity of the increase of suicide with increase in movement of the wind is too marked to allow any other theory than that of a causal nexus. This effect seems to be much greater upon the suicide than upon any of the offenders mentioned in the study cited. It is, however, shown to be as great or even greater for all classes of crime in the Colorado climate, where wind is an important factor in the production of high electrical states. The other study, however, showed very slight wind effects for New York City, and their comparison with

PSM V58 D622 Suicide and wind correlation chart.png

Fig. 6.

this would seem to prove that the mental states of the suicide and of the street brawler are very differently influenced by it.

It is difficult, in conclusion, to summarize the results of this study in such a manner as to be of much value or to bring forward theories which are certain of any long tenure of life. The whole method of the study is too new and untried, and the number of data inadequate. The bare facts revealed in the preceding paragraphs must prove of much more value than any hypothesis drawn from them at this stage of the investigation. Still, there are a few generalizations which seem worth noting, especially as they are based in part upon findings which are entirely contradictory to popular opinion with regard to the time chosen by the suicide for the final act.

The first is that suicide is excessive under those conditions of weather which are generally considered most exhilarating and delightful—that is, the later spring months and upon clear, dry days. Reference to Figs. 1 and 2 proves this conclusively for the number of data and the locality studied. It was also noted that there were the greatest numerical excesses for the most agreeable temperatures. Barometrical conditions can hardly be referred to the categories agreeable and disagreeable, but for humidity and wind the relation will hardly hold, since we have the greatest excesses during high humidities and great wind velocities, both of which are unpleasant. Yet these facts would not invalidate our first statement, for neither high winds nor great humidities bring a scowl upon the face of Nature that can be compared with that of a wet, drizzling day. In fact, a day may be bright, and be both windy and humid. Yet these latter conditions have effects peculiarly their own, as shown conclusively by the study of deportment already cited. They are, for wind, the production of a neurotic condition in which self-control is in a marked degree lessened, and for high humidities the production of a minimum of vital energy. The former is shown especially in the study of the school children, and the latter of the death rate. These facts make it possible for us to amend our statement that suicides are excessive during the most noticeably delightful conditions, by adding: coupled with especially devitalizing ones.

But this does not in any way account for the seemingly anomalous effect of bright weather. To me the only plausible hypothesis is that of contrast. Investigation has seemed to prove that very few suicides are committed on the 'spur of the moment.' The act is generally premeditated, and its consummation deferred, sometimes again and again. We can hardly doubt, either, that it is dreaded, and the hope entertained, even to the end, that it may not need to be. During the winter months that hope must be centred on the belief that when Nature smiles with the spring sunshine all will be well; on the gloomy day, when the morrow comes with its exhilarating brightness, the present cloud of unhappiness will be gone. The love of life is still strong, and the grave can not be sought while there is still hope for better things.

But spring comes with all its excess of life, and the morrow with its brightness, but do not bring to the poor unfortunate, unable to react to these forces as of yore, the hoped-for relief. He thinks of other springs when the bluebirds sang happier songs, and of other sunshine which had set his blood tingling. The drowning man had waited long for the straw; it came and he clutched it, but it sank beneath his weight.

  1. See 'Conduct and the Weather,' Monograph Supplement No. 10, 'The Psychological Review.'
  2. See 'Conduct and the weather'