The engine-driver is told that red is a "danger" signal, green a "caution" signal, and white an "all right" signal. Therefore, it is very necessary that he should know what is meant by these colours.

It must be noticed that it is on account of there being approximate and absolute psycho-physical units that such great difficulty has been found in constructing an adequate test for colour-blindness, as it is the approximate, and not the absolute psycho-physical units of which we wish to know the number. It will be seen that it is not merely a matter of shade as far as the colour-blind are concerned, but a distinct difference in tint. The normal-sighted could divide the green of the spectrum into yellow-green, green, and blue-green; and would, in the majority of cases, be able to range all greens under these three classes. As the two-unit colour-blind see two colours only, they must name colours in this way.

We wish to exclude all those individuals who are included in the following three classes :

I. Those who possess a pyscho-physical colour-perception with three or less units.

II. Those who, whilst being able to perceive a greater number of units than three, have the red end of the spectrum shortened to a degree incompatible with their recognition of a red light at a distance of two miles.

III. Those who are affected with central scotoma for red or green.

I will now explain why these three classes of persons should be excluded.

With regard to the first class. three-unit, the two-unit, and the with the facts previously stated.

This class includes the one-unit, in accordance The three-unit never,

under ordinary circumstances, mistake green for red, but confuse green, purple, and gray. Colour is a feeble quality

of objects to them, and nervousness or excitement may reduce them to the condition of the two-unit colourblind.

The two-unit regard green and red as almost but not quite identical, and this fact is one which it is nearly impossible to make a person, who has not thoroughly studied colour-blindness, comprehend—either the colour-blind himself, the public, or an unqualified examiner. They find that many colour-blind persons are able to recognize different colours, and correctly name them, and therefore set down the mistakes made to want of education in colours. It is astonishing with what accuracy many colour-blind (two-unit) persons name colours. I have met with several who were nearly always correct when they named a colour. One educated adult, in particular, had become so expert that he was often able to baffle attempts made by his friends to show that he was colour-blind. He was well aware of the fact of his being colour-blind, though, when I asked him the names of various coloured objects, he was nearly always correct, and did not content himself with using the ordinary colour-names, but employed such terms as cerise. He told me the method he adopted; he said all colours appeared modifications of blue and yellow. The brightest and purest yellow was yellow; slightly darker and not so pure, green; darker still, red. The brightest and most typical blue, violet; less bright, blue; blue, with a tinge of gray (dirty blue), purple; very impure blue, cerise. This colour appeared to him blue by day, yellow by gaslight.

We cannot wonder at the unfavourable comments of unqualified persons, when we consider that the large majority of recorded cases have been of school children and uneducated persons. This source of error I eliminated by using for my standard cases and the following ex

periments only educated persons who had tried to train their colour-sense, and were well aware of the names of colours.

The following will show how it is that the colour-blind. are able, under ordinary circumstances, to distinguish between the colours included in one of their units. All colours have not a similar degree of luminosity; thus, yellow is much the brightest colour. To the two-unit colour-blind, red, yellow, and green, have, as far as colour is concerned, a very similar appearance. They are not exactly alike in colour, because they are included in an approximate, not an absolute psycho-physical unit. Green looks a lighter and grayer colour than red. A normalsighted person might be given a bundle of wools, consisting of three kinds; the first different tints of yellow, the second tints of yellow mixed with gray, the third shades of yellow-that is, yellow mixed with black. He would be able to arrange these in three groups with few mistakes.

With regard to the second class. It is very important that persons belonging to this class should be excluded, and yet none of the ordinarily used tests detect them. The rays of red at the extreme left of the spectrum are the most penetrating, as may be seen by looking at a light or the sun on a foggy day, or through several thicknesses of neutral glass. It is chiefly by these rays that we recognize a red light at a distance; and it is therefore of great importance that a sailor or engine-driver should be able to perceive them.

With regard to the third class. This condition is one in which a person might be able to distinguish colours easily when they are close to him, but fail to distinguish them at a distance, owing to the nerve-fibres supplying the central portion of his retina being impaired. As a light

at a distance occupies the central portion of the visual field, it is essential that the corresponding portion of the retina should be normal.

We also do not wish to exclude persons who, though partially colour-blind, have a colour-perception sufficient for all practical purposes.

With regard to the test to be used. If the persons to be tested have to distinguish between the standard red and green lights, these lights should be used as the basis. of the test; because, if any other test were used, we should still have the same problem before us, from a practical point of view. A sailor might (with reason) object to any other test, and say that because he cannot distinguish between a green and a gray wool, it is no reason why he should be unable to distinguish between the red and green lights.

The candidates' capacity in this respect may be tested with a Lantern, which is described below.

Though the fitness of a candidate could be determined with the Lantern Test, the examiner, especially if inexperienced, would, in many cases, be in doubt as to whether a candidate had made a mistake through a slip, or from a definite colour defect. This is especially liable to occur when the candidate is slightly colour-blind, and therefore, in order to save time and aid the examiner in detecting colour-blind persons with certainty, I have constructed another test, which I have called the Classification Test, and which may be used as a preliminary to the examination with the Lantern Test.

A conscientious examiner will then have his diagnosis confirmed when the results of both tests agree, and when they differ (I have not yet come across a case in which they have) he will be able to go more thoroughly into the

case.

I.-REPRESENTS THE LANTERN TEST.

A shows the lantern as presented to the candidate, a slide being in situ.

B shows a side view of the lantern with three slides in situ, a being a coloured glass, b and c modifying glasses, d is the front view of a slide, e the pile of thirteen slides.

1, 2, 3 and 4 are the test colours-Orange, Violet, Red and Blue-green; a indicates the wools, b the silks, c the pieces of coloured cardboard, and d the squares of glass, forming the confusion colours of the test.