patient using the shutters until he is satisfied that he has obtained the correct junction.

Question-If you had say five points of difference, would they always come exactly in the same place?—Yes; such a case would not see orange as a definite colour.

Question-Would a patient recognise no orange, supposing a single colour was suitably chosen and distributed over a sensible field, and beginning from the sodium line?-A 5-unit case usually objects to the term orange; he would probably call it reddishyellow.

I have not come across any 1-unit cases; that is the only one on the diagram that is not drawn from my own experience. I should explain that the diagrams are all drawn the same lengths, to demonstrate the psychophysical diminution of colour-perception and not the shortening of the spectrum, in which case another effect is produced, viz., the junctions of all colours are altered; a 2-unit, with shortening of the red end of the spectrum, puts the junction of his two colours nearer the blue than a 2-unit with an unshortened spectrum.

Diagram No. 6 shows a transition from red to violet; in such cases there is no neutral band in any part of the spectrum. The one colour passes into the other without any definite intermediate point.

In examining for scientific purposes the spectrum would be the first test, afterwards wools, pigments, or lights, in accordance with the spectrum examination.

Question-Would it not be easy to coach a person for this?— No, because he would never be able to hit off the exact junctions of the colours.

Question-Which class of the colour-blind would you consider as representing dangerous cases for signalling purposes?—I. Those who possess a psychophysical 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. The 3-unit would be unsafe, for though he would always recognise red and green, even to the lowest degree of luminosity, he would confuse yellows, especially dark yellows, with reds and greens, and generally call them reddish-greens; in fact, yellow has been described by such a patient as being of the same colour as a red clover field in full blossom. The 2-unit cases and below are absolutely dangerous. The 5-unit and 4-unit are safe.

[The Witness here handed in his Pocket Test, and explained its use, Captain Abney being asked to pick out all the shades of orange.]

Question-Would you describe a person as not having distinct orange perception, who could not mark out the definite regions on the spectrum bounded on one side by yellow and on the

other by red?—Yes; I should describe such a case as not seeing orange.

Question I understand Dr. Brown's test comes out under your patronage; may we therefore take it that you approve of that method?-Not at all. A medical man might roughly test with it.

Question-Do you test by nomenclature or matching?-By nomenclature, combined with matching. Many normal sighted persons fail with Holmgren's test because they think a shade is a colour, paying as much attention to the one as the other; if you say "I want you to pick out all the greens," you give them something tangible to go upon. If a person in picking out twenty or thirty greens also picked out half-a-dozen reds, it would be certain he was colour-blind; but if he has to match a green wool he might pick out with the greens a light brown, and not be colour-blind at all, and the error could be rectified by explaining to him that the colour he selected was greenishbrown, reddish-brown, or yellow-brown, as the case might be. With Holmgren's test, under the same circumstances, he would have failed. This might be confirmed by asking him to classify the whole 150 colours.

Question-Do I understand you that a normal sighted person might pick out brown instead of pale green ?-Yes, because he might pay more attention to shade than to colour. In testing practically I should first use the Classification Test. I should not begin with the spectrum with a practical test; it would not be convenient, and persons would object to it.

[At this point the witness was asked to apply the Classification Test to Mr. Rix. Mr. Rix was first requested to pick out all the shades of orange he could see, and in so doing he selected two skeins of wool of a decided light green. Dark blue and violet were matched as being of the same colour. In matching reds, a reddish-brown was picked out, but described as having more blue in it, and blue-green was sorted with the drabs, and referred to as being brighter.]

WITNESS.-The Classification Test is used in order that inexperienced examiners might not have to depend upon the Lantern Test-in which not more than twenty answers are required. I do not use the Board of Trade colours with that test, but my own. The person examined should be able to distinguish between the red, green, and white lights, either alone or modified with the neutral glasses.

Question Why do you think it necessary to have a preliminary test to this?-In order that au inexperienced examiner may feel certain that the mistakes made with the Lantern Test are not due to colour ignorance.

Question-Have your investigations been with pathological or congenital colour-blind cases?-With both.

The principles upon which I examine are as follows:-The first principle which guided me in the selection of colours may be

illustrated in the following way :-Let us take an ordinary 2-unit colour-blind, and, having given him the set of wools belonging to the Classification Test, ask him to pick out all the reds. On examining the pile of wools selected as red, it will be found that the majority are red, but in addition there will be some browns, and yellow-greens. If he be then told to pick out the whole of the greens, the greater number of those selected will be greens, but there will be also greys, browns, and reds. In each case it will be seen that the majority of wools are of the desired colour. If another 2-unit colour-blind be examined in the same way it will be found that, though he may not make exactly the same mistakes, he will in all probability pick out the same greens to put with the reds, and the same reds to put with the greens. The same result will be obtained if the colour-blind persons be asked to name a large number of colours. They will in most cases name the colour correctly. It will be noticed that the greens which were put with the reds when classifying the colours, will be called red in naming them. It is evident that the same idea has guided the colour-blind in each case.

This shows that, though a person may be red-green blind, he is not absolutely red-green blind in the sense of being totally unable to distinguish between the two colours. This is what we should expect, as the red and green are included in an approximate, not in an absolute psychophysical unit. The fact that they are actually judging by colour may be demonstrated by giving them coloured materials of different kinds, or by asking them to name a large number of coloured objects. To a person with a spectrum of normal length and no neutral band in the blue-green, it is necessary that the colours, to be considered as identical, must be included in an absolute psychophysical unit. One of the most definite signs that persons with a neutral band in the bluegreen have a more defective colour-perception than the ordinary 2-unit, is that they will put together as identical a red and green which are distinguished by the ordinary 2-unit. In addition to this, they will mistake the reds and greens which have been confused by the ordinary 2-unit.

It will be seen that if we take a 2-unit and ask him to name a number of red and green wools, in the majority of instances he will name them correctly. But as, almost invariably, the same wools are chosen, for all practical purposes the same result would be obtained by asking a person to name a few of these wools. What more decided and brighter greens could we have than Nos. 76 and 94 of my Pocket Test? yet these are two of the greens which are called reds by the 2-unit. We should have accomplished as much by asking a colour-blind person to name Nos. 76 and 94 as if we had asked him to name a large number of greens. The colours in a test should, therefore, be those which the colour-blind are particularly liable to miscall. At the same time, their nature should be unmistakable to the normal sighted. My second principle is that a colour-blind person will name

colours in accordance with his psychophysical colour-perception, and thus show distinctly to which class he belongs.

The third principle is that colours may be changed to colourblind persons whilst leaving them unaltered to the normal sighted.

Fourthly, the phenomena of simultaneous contrast are much more marked with colour-blind than with normal sighted persons. Two colours not changed to the normal sighted, on being contrasted, apparently alter considerably to the colour-blind.

These tests are described in full in my book on Colour-blindness and Colour-perception in the International Scientific Series.

Question-How would you proceed, supposing you were asked by a railway company to test 500 men? What arrangements would you make, and how long a time would the examination be likely to occupy?—I should examine each separately, taking care that the others did not look on. I should first examine with the Classification Test, and then put them through the Lantern Test, taking twenty answers in each case. The process would only take about five minutes for each man, with even one examiner, because one man could be going on with the classification whilst another was being examined with the lantern. I allow ten minutes for each man, because I think that it is a great mistake to hurry, or be in any doubt about a case.

Question-At that rate it would require 80 or 90 hours to examine the 500 men?—It would mean a considerable expenditure of time. You want to know if a man can distinguish between red, green, and white at a distance of two miles. You might commit yourself to the Lantern Test alone; but if one man was sorting the wools while another was being tested with the lantern, it would take little or no longer to employ both tests, and in rejecting a man the double test would be conclusive. I regard both tests as desirable, and the Lantern Test as essential, for that would detect scotoma, whereas the Classification Test would not.

Question-You said you tried a progressive atrophy case, and that he was a 2-unit man, and his junction at about the E line?— I cannot say without referring to my note-book, but he saw one part of the spectrum as whitish and the other blue, only seeing those two colours. [Capt. Abney said:-When I tried him the junction of the white was at X 4.733, between F and G, and nearer to F, there being a sudden commencement of blue at this point. At 26.5 of my scale he saw a little blue, and at 26.75 no colour.]

The essential part of my theory is that psychophysical perception is due to the brain and not to the retina. The theory I have formed is that the visual purple is liberated from the rods by light, and forms a photograph at the back of the retina, and that the cones only act as transferring organs of the percipient fibres, transferring the impression of the photograph to the brain.

he

Question-Is not that something like Kühne's theory?—No; took a different view, the objection to which was that there are no rods in the yellow spot; but according to the theory I have advanced it would be essential that there should be no rods at the yellow spot.

Evidence of Capt. ANGOVE.

The CHAIRMAN. You are, I believe, the Captain Superintendent of the Peninsular and Oriental Steam Navigation Company?Yes, the Marine Superintendent.

You are probably aware that this Committee is investigating the question of colour-blindness, and the precautions taken by steam-shipping companies and railway companies against accident from this cause; we are therefore anxious to hear what is the practice of your company with regard to examining officers ?One of the Managing Directors and myself first examine applicants with regard to objects at different distances from the windows, then the Company's Medical Adviser examines them with coloured wools and for distance sight. They have to arrange and name the different colours. The vision for distance is tested by placing the candidate at the regulation distance from ophthalmic large test types, and the near vision by corresponding small type. It is required that both eyes should be equal to the average of good sight. We find, I am sorry to say, a great number who are not up to the standard, and often have to reject candidates in

consequence.

Question-That is the procedure for candidates for the posts of officers?—Yes, but there is also a Board of Trade examination before coming to us which they have to pass in getting their certificates for various grades of officers.

Question-Do you take pupils in your Company ?-No, not now; in former years we had two training ships, but they have been discontinued, so that every officer must now hold the Board of Trade certificate, which includes colour-vision. We often find candidates deficient, and reject a great number for defective sight.

Question-Then that points to the Board of Trade examination being unsatisfactory ?-Yes, I do not think their examination is sufficiently rigid.

Question-Would you in your examinations reject candidates both for colour-vision and form?-Yes; and we give equal attention to weakness of sight with regard to seeing at long distances.

Question-Do you give rejected candidates a second opportunity?—No, but some go to an oculist's on their own account, and we get a special report from him. The weakness may be only of a temporary nature. I attribute a great number of cases to over-smoking with young men. We have traced several