Popular Science Monthly/Volume 72/March 1908/Railway Accidents and the Color Sense




IN considering whether some of our frequent railway accidents may not be due to the character of the signals we employ, it should be borne in mind that these signals often must be caught and instantly translated into action under conditions of uncommon mental stress. And for this reason, defects of the symbols which might otherwise be far from serious do now become of vital moment. Yet it has been said that the work of the locomotive engineer seems to the observer more difficult than it is—that the long training through which these men must pass permits them to carry lightly their great responsibilities. It was the more interesting, therefore, when, on an express-engine not long ago, we had come to the end of our long course, and the din and jostle had given way to calm, to hear the engineer speak of the tension of his work. He had been at the throttle but three hours that day, and after going for a time to the round-house, would take his express back over the same run that night. "My partner," said he, "will have the run to-morrow. No man could stand it, holding her down in this way day after day." And so the engine crews on such a swift express lie off alternate days, and the engineer and fireman may not take out their train unless the entire preceding day has been a day of rest. Such carefulness on the part of a great corporation calls for praise which should be all the less restrained when so much must be said to-day of the shortcomings of our railways. Yet there could hardly be stronger proof of the strain under which the engineer must labor; for no company would give to its hardy servants every alternate day for freedom, unless experience had taught that the service itself required it.

Nor is it difficult to appreciate in some measure the severity of the work. Various duties that on an ocean steamer are distributed among helmsman, lookout, engineer, and the officer on the bridge, here fall chiefly upon a single man, and this where the care and instant judgment required seem at times to be not far below those needed for the guidance of a ship. The locomotive engineer must control a marvelously complex and ponderous piece of mechanism, keeping his sight and hearing and sense of shock so alive that amid the universe of whirl and glare and explosive rattle in which, for the time, he is centered, he can detect the foreign note or quiver that speaks of disarrangement. He must know that his outside lights are burning bright, that the water in the boiler is sufficient, that the air-brakes are in perfect working. He must from moment to moment glance at the hands of his watch, and must know exactly where he is upon the road. And yet all the while his eyes must hardly be taken from the darkness into which his engine rushes, to catch the first glimmer of the signal which is his guide.

Since the safety of many lives thus depends upon these signal lights and upon their sudden clearness to a mind that must attend to many things at once, the symbols should at all times be the least ambiguous that can be planned. Yet the present night-signals, given by colored lights beside the track—upon many roads, white for "safety," red for "danger," and green for "proceed with caution"—are open to grave objections. For the human eye at its best and without abnormality is liable to mistake the signal hues at night, especially when the outward conditions are anywise untoward, whether by the distance or the low-burning of the lamp, or by fog or smoke or storm. And even when the colors are perceived with perfect accuracy, the use of the common oil-light called "white," as one of the signal colors, throws a dangerous task upon the engineer, inasmuch as it requires him to take constant heed lest he regard some window-lamp, or other meaningless light along his course, as a sign that all is well, and in consequence rush onward to his train's destruction.

That objections of this character are supported by strong evidence, and are not of merely theoretical importance, but are connected with actual and known disasters—to make this clearer is one of the main purposes of the present paper.

In regard to the use of white as part of the signal code at night, the danger from this source has long been recognized by leading signal engineers, although in spite of this recognition its use continues on a large number of our American roads. It is not many years ago that an accident occurred at Whittenton Junction, Massachusetts, from this very cause. The engineer mistook a lantern hanging from the gate at a street-crossing for his safety signal, and crashed into another train. More recently, Mr. Baggett, of the Galveston, Harrisburg and San Antonio road, has given an instance where disaster resulted from the use of white. The railway signal, in this case, was exactly in line with a light shining from a high bay-window; and when the signal light itself one night was out, the engineer mistook the light in the window for his signal, and a serious accident was the outcome. And other cases are reported by the Interstate Commerce Commission. A switch light happened one night to be extinguished, and the engineman "failed to notice the absence of the light, being deceived, he says, by lights in the vicinity"—a deception which brought damage amounting to $3,500, a maiming and a death. On another occasion an engineman runs past a signal which the station agent declares was set at "stop," but which the engineer himself asserts was showing "clear." Immediately after the passage of the train, the signal light was found to have been extinguished, as it had been once before that evening; it is not improbable, therefore, that the "clear" light which the engineer saw was some neighboring light which he took to be his signal. His mistake brought death to 18 persons, injury to 57 more and a loss of $15,720.

With this evidence of the danger which lies in using white for color signaling, and before passing on to consider red, a word should be said of green—a color which, partly by its strong contrast with red, its companion in the system, has found most wide acceptance. Green stands out distinct from the common lights of street and house; it readily makes an impression upon the normal eye. But persons who are not color blind are liable to be weak in their sense of this very color. And smoke, one of the great disturbers of signals on the railway, has a serious influence upon green. For smoke, which makes white or yellow lights look red, does this by making ineffectual the very rays that are so important for giving a green light its greenish cast. The simple experiment of holding a smoked glass before a green railway-light will easily show how hostile smoke is to the passage of green rays. Now when, for this or any other reason, green comes dimly to the eye, especially when sight has grown accustomed to the dark, it has the misfortune of appearing, not green at all, but a pale and ambiguous light that is indistinguishable from white. Under such circumstances, especially upon those roads where both white and green are signal colors, the danger of their confusion is not imaginary; nor is the danger of green's total obscuration slight. In the records of the Interstate Commerce Commission occur more than one instance where the failure to observe at night a "distant" or "caution" signal, which is often green, has been an important part of the cause of fatal accidents.

But after all, the core of the present system is red, and to this our main attention should be given. The color is usually obtained by bringing before the semaphore lamp a glass which, acting like a filter, permits the passage of those rays that are red or reddish, and holds back from the eye all other light. Such a ruby glass, by killing off in this way all that portion of the flame's light which is green or blue or violet, and often all that is yellow, does of necessity greatly reduce the brightness of the signal, leaving it in many cases about one fifth as intense as when, by the signal mechanism, the red glass is removed from the front of the lamp. This readily explains—what any one can observe—that in a cluster of signal lights equally remote, the white signals normally outshine to a marked degree the neighboring signals that are red. This, of itself, is an undesirable condition, since the sign of danger should of all be most outspoken.

The disadvantage under which the red danger signal labors is, however, quite insufficiently expressed by saying that the ruby glass often virtually destroys fully four fifths of the light from a lantern flame already none too bright, and to this extent increases the liability that the most momentous of the signals will at some crisis be seen too late or not at all. Even the remaining portion is often far less effectual upon the eye than its physical quantity would lead us to expect. The importance of the matter for signaling will perhaps justify some further account.

If, by reliable devices of the laboratory, a semaphore light showing "white" be gradually reduced in brightness, a point can easily be found where the eye, grown accustomed to the dark, can just perceive the light. And when for comparison a railway ruby glass, or "roundel," is placed before the lamp, the observer now obtains no conscious impression at all. But instead of having to increase fivefold the brightness coming through the glass (as one might expect, knowing that the red glass is pervious, say, to but one fifth of the light of the flame), it is necessary to increase it no less than fourteenfold. Such an increase is the least I have found necessary when experimenting at night over a stretch of more than four thousand feet and when smoke gave a relative advantage to the red. Within the laboratory the red has never been perceptible until the light was increased eighteen times the brightness required for white. Such, however, are the most favorable experiments, and are by no means average ones. On the average it is necessary to increase the light as much as thirty times before any conscious impression at all is made by the light through the red glass. One of the subjects of this experiment—a man who would pass the usual tests for color-blindness—has still remained insensible to the red when the light is increased to seventy times what is needed for the white! Such facts as these show clearly that by merely looking at a cluster of railway signals, or even by taking the usual tests of their relative intensity or visibility when shining bright, we get no adequate idea whatever of the difficulty which the eye has with very feeble reds. And feeble reds are no great rarity in the actual conduct of trains. The many influences which render signal lights obscure thus act with a peculiarly fatal force upon the very color which is our chief reliance for the protection of life.

And the doubt thus raised regarding red is not allayed by the reports of railway accidents. For in these reports the frequency with which engineers fail to observe red signals at night is a most impressive fact. It is often impossible to tell assuredly why men at such times are unconscious of the danger sign; but even when allowance is made for the sheer carelessness of trainmen, or for exhaustion resulting in their sleeping while still on duty, there seem to be cases enough to warrant some suspicion of the virtue of the danger signal itself. In the summer of 1904 an engineer at night ran past signals at "stop," and into an open drawbridge with his train. About a year later, a disaster was caused by the failure of an engineer at night to heed two block signals and a flagman, the engineer himself losing his life by his mistake. On still another occasion, a train ran past three or four warning red lights, and by a collision with a passenger train ahead brought death to 23 persons, injury to 85 and a property loss of $7,000. And the inherent difficulty of perceiving weak red may have been a contributing circumstance to that wandering of the attention of an engineer who recently, after passing a "distant" signal obscured by smoke, failed to notice "until he was quite near it" his "home" signal telling him to stop, and crashed into the rear of a passenger-train. This failure to see in time the warning light cost seven lives, brought injury to no less than 142 persons and destroyed property amounting to $44,000. And finally, on the night of December 30, 1906, a collision occurred in the District of Columbia, due in part to the failure of an engineer to see a red light obscured by fog; and in this disaster 43 persons were killed, 63 were injured and property valued at $16,000 was destroyed.

These fearful results, depending, as they do, upon the failure to see red lights, dimmish greatly one's confidence in this color. And this confidence grows still less as we bear in mind the many instances, which have not been adduced at all, of utter failure to observe the red "tail lights" of trains, red switch lights, or red hand lanterns—failures that ended in deadly accident. The insufficiency of a red hand-lantern carried down the track for the protection of a standing train is doubtless in part to be ascribed to the unexpected place in which these signals must of necessity be shown, and the engineers' unreadiness to note them; but it is not improbable that the very color of the signal contributes to its failure. We are accustomed to think of red as exceptionally impressive; and it truly is in many respects an effective light, attracting the attention when once the eye catches it in strength. But at degrees of illumination that would be ample for some of the other colors, it cease to penetrate the mind—somewhat as in photography, red of all colors has least effect upon the sensitive plate. Red may some day come to be regarded as a danger signal, with an unusual meaning to the words. And yet, taking all things into consideration, it is perhaps the best color to use as a sign of danger, if color must be used. Such a conclusion, however, reveals but too clearly the weakness of a system based on color—reveals how fatally mistaken it is to make the life and safety of passengers dependent upon the hurried recognition of colors at night by any man, above all by a man who has many and most insistent duties besides.

That the color-sense is wholly unfit for the office it holds in rail-roading is hardly open to any doubt whatever. One must speak with less assurance, however, as to what should take its place. But even here the general principle that might guide the change is reasonably clear. Our eyesight detects two different features in objects—their color and their spatial character, such as shape, position and movement; and the sense of color is far less primitive and vital and masculine than is the rude sense of space. Nature seems to have held the sensitivity to color a cheap and slighted accomplishment, to be crowded out or postponed to the mere finishing school, like young ladies' French and dancing. But the rugged feeling for place and direction is early given and pressed deep until it becomes a central fact in self-preservation and advance.

Now if the eyesight of the engineer is to be depended upon at all, it is this more fundamental and stable portion of it that should be given responsible work. If the practical difficulties could easily be met, the power to distinguish between rest and rapid movement of some conspicuous object would be the best to call upon in signaling. For we, like the bird and beast in the woods, are alive to slight quick movements in the field of view, far more than to color or even to shape and size. When the arms are waved or a lantern swung by hand to attract attention, appeal is instinctively made to this deep and primal interest in moving things. But next to this, the simplest and least erring of our visual perceptions is of large differences like that between a vertical and a horizontal line or one aslant. Now these rough and simple elements are precisely those used for the day signals of most block systems, where there is an extended arm placed high beside the track, and its direction of pointing—up or down or at some angle intermediate to these—tells the engineer whether the track ahead is open to him or closed or to be entered only with caution. Such signals make no prime appeal whatever to the sense of hue, but only to the sober feeling for visual place. And there seems to be nothing to prevent that this same principle of signaling should be carried over into the night and be even more successful there. For the extended vane used for the day signal often is before some unpropitious background of buildings or of trees against which it stands out in no strong relief. But at night it would be possible to use some self-luminous line of light that would appear sharp and unmistakable against the dark.

The detailed mode of applying such a general principle belongs to mechanical art rather than to psychology. But lest the principle itself should be misjudged for want of some more definite form in the mind, it might be well to imagine a row of incandescent lights inserted in the signal-arm now used by day, but lengthened and otherwise modified in whatever way its new work might require. Considerable intervals may be between the lights, and yet from a distance they will seem a continuous line; and as for length, experiment both in the laboratory and over a distant stretch at night shows that the main directions of such a line can be caught by the normal eye when the length is about a thousandth of the distance from which it is to be read. Three or four lights in a row about five feet in length would thus suffice for giving an engineer his signal a mile before he passed the post. A space signal given in some such way would naturally require more feeding of electricity or gas or oil—than does the single small wick flame that now gives forth the colored beam. But on the whole it would perhaps be well to spend fuel and light rather than life.

The mere imaging of a generous and glowing line to give the signal, will at once quiet some grumbling doubts that come from failures hitherto. Certain older attempts to use the space principle for railway signals at night could easily bring misgiving if the real cause of the failure had lain in the space sense itself. But, in fact, there has been no readiness to use the amount and extent of light needed for a proper signal. One of our roads tried to guide its trains by means of two lights whose changing position with reference to each other should give the sign to the engineer; and this signal was found unsatisfactory because these two lights blended into one when looked at far away. But for the most part the attempts have been confined to spreading out into a band, by means of a reflector, the light of a single lamp—a band of light that was too faint to be well seen when the reflector became dimmed by smoke or the corrosion of the weather. Such crude attempts to make a spatial signal were of course foredoomed to failure, and give no reason to distrust the perception of space itself. They simply prove that sufficient brilliancy of light must be maintained, and that this brilliancy must be stretched to sufficient length—conditions which can certainly be fulfilled at least wherever there are electric lights.

One can speak with greater confidence because of the practical success of such spatial signals in another field. The use of the luminous line is already well established in the navy. Two movable arms, each provided with a row of incandescent lights, here rapidly convey, by the direction in which they point, their message from ship to ship, or to the shore. And even with comparatively short lines of light, their position is legible by the unaided eye at a considerable distance. The outcome here gives ample reason to believe that it would be possible to apply the same general method to the railway service.

The advantages of relying on our space-perception instead of on the color sense will probably in time be recognized as far outweighing whatever difficulties there may be in the change. The new plan would merely require us to suppress the worse half of the present composite set of block signals—the half, which relies on color—and to render universal its better portion which already signals by direction. The present system would thus be simplified and fulfilled, rather that annulled, and there would be no need of training engine drivers to an unfamiliar code. And while the perception of the trend of a line of light requires that the refractive power of the eye shall be normal or shall be corrected by the use of glasses, the engineer's work even now demands that his spatial vision shall be keen, and thus no innovation would be made.

But even were there many objections to the use of spatial signals, they must be grave indeed to outbalance the fact that a line of light not only frees us from the treachery of the color sense, but gives a symbol that is distinct from the usual lights of the window or the street, and at a stroke renders well-night impossible those accidents that come from mistaking foreign lights for block signals. Moreover, we should then have a system wherein danger would be indicated at all times as clearly and as unmistakably as safety, whereas in the present code the red danger signal can too readily remain unseen. An important advantage besides would be that signals of the kind here proposed could hardly, by influence of smoke, or fog, or storm, be made to seem the very contrary of what they really were. A green light may look whitish, or a yellow light red, by mere conditions of the air. But a vertical line can not well be made to appear horizontal, or a horizontal diagonal, by smoke or fog. Its message might be cut off entirely, but could not readily be distorted into its fatal opposite. In this, as in so many other ways, a change of usage commends itself to the critical sense.

In urging that we no longer rely upon the color faculty for the safety of our trains, I have spoken almost exclusively of those difficulties which color offers to eyes that are entirely normal and sound. And upon such facts the main objection to the present system may well be based; for they are strong enough in themselves to condemn our usage and to demand that it be changed. But the reasons so far given are immeasurably strengthened by the existence of color-blindness and other defects of the sense of color. There are some men, it is true, who believe that the danger from this source is entirely averted by the current examination of engineers. No one would wish needlessly to lessen faith in such examinations. And yet it should be more widely known that defects of color vision are not always easy for physicians to detect; far less are they for laymen. And since many dangerous cases are known to slip through the meshes of medical examiners elsewhere, it is reasonably certain that the same is true with us. Dr. Stadfeldt, of Copenhagen, in a recent examination of 295 pilots (who, like enginemen, must distinguish colored signals), found 17 who were defective in their color sense. And Professor Nagel, of the University of Berlin, one of the great authorities in this department of research, and who has been called upon to assist the Royal Prussian Railways, has lately found in responsible positions like those of engine driver, fireman, switch tender, no fewer than twelve typical instances of red-green blindness among men whose sense of color had been officially tested and approved four or five times. Of about 300 employees of all branches of the service, all of whom had been tested at least once—and almost all of them more than once, by physicians and not by mere laymen—Nagel reports five per cent. to be typically color-blind; not color-weak merely, but actually color-blind. It is, therefore, difficult to partake of that happy confidence expressed by one of our leading railway journals, that "the railroads have long since done away with the dangers of color-blindness, by taking color-blind men off from their engines." We must, on the contrary, believe that the undiscovered presence of color-blind and color-weak men upon our engines adds to the many reasons for refusing longer to intrust the safety and life of thousands to one of the most fickle of our human faculties.