]\gt ; fixed width .
Light of sensibly one wave-length , i.e. monochromatic light , * ?
will pass this slit , and by meaus of a lens placed in the beam of this light an image of the first face of the prism which is used to form the spectrum can be formed on a screen .
In this way a monochromatic patch of light is obtained , the brightness of which depends on the nature of the somce of light , the width of the collimator slit , the width of the slit placed in the spectrum , which for short will be called the moyable slit , and the dimensions of the lenses employed .
Further , if alongside this coloured patch is formed a white patch of light produced by light which proceeds from the same source but has not undergone dispersion , and that by some means or other the intensity*of this white light is altered till the coloured and white patches appear to the eye equally bright , then the intensity of the white light , ured in any arbitrary units , measures the luminosity of the light of that colour which is passing through the movable slit .
Since the unlt in which the white light is measured is arbitrary , we are not concerned * The physical brightness of a light , i.e. the stimulus , will be spoken of as the ntensity , the term being rved for the sensation produced when the light nters the eye .
Luminosity Curves of Colour Vision .
405 with the absolute intensity of illumination of the white patch , and may use screen satio ouantity oightany e given person tement ouminosity ogiven cight iemployed iifferent experiments.urther bbserved t patch appears tctrum iomparison orightness olouredhim wrightness obite patch a appears timThe above definition of what is meant by the luminosity of a colour in the spectrum of a given source is equivalent to that employed by Sir William Abney in all his work , and the only reason for , it is that to follow the S reasoning used in this paper it is essential that this definition should be kept in mind .
A convenient arrangement for conducting measurements of empl ebservati iveIlluminosit hescribed bFesti eThis iardly tlace tiscuss tccuracy which w :judge of the comparative brightness of , say , awhite and agreen light .
It is admittedly a difficult operation , but the results which are quoted below show that if the conditions are suitably chosen the measurement can be performed with considerable consistency .
The difficulty of comparing the obviated iflicker method employed , hich toloured ahitelights auccessively occupy tcreen ahebrightness ocoloured light wwhite light ilmost completely intensity of the white is altered till the flicker vanishes .
If suitable precautions are taken even quite inexperienced observers are found to be capable of making quite consistent measurements by the flicker method , and a satisfactory piece of apparatus for such measurements is described later .
If the movable slit described above is placed in different parts of the spectrum and the intensity of the white which appears equally bright is measured in each place , on plotting these intensities as ordinates , the abscissae prism deviations , a luminosity curve is obtained , that is a curve which gives the luminosity of the different colours in the spectrum of the source and for the particular prism employed .
As has been mentioned , the unit in which the intensity of the white is measured is quite arbitrary , and throughout this paper it is taken as such that for a eye the maximum luminosity in the spectrum of the brightness employed is The abscissae of the luminosiby curves given are also 'Phil .
Trans 1886 , Part II , p. 455 .
The intensity of illumination of the screen corresponding to 100 is throughout about , or candle-feet .
been shown ) Abney , Tufts , Ives and others*that , etc. *This statement requires a certain amount of limitation , and must be held only to apply to spectra of moderate brightness and to sources of light such that the light is not ?
chiefly due to the blue and violet .
The proposition that the luminosity of the sum of two lights is equal to the sum of the separate luminosities does not apply to very bright : lights , and in the case of very small intensities may not apply in the blue and violet .
having Normal and .
407 That is , the luminosity of the recombined spectrum is equal to sum of the luminosities of its parts .
, Now the sum , etc. , is proportional to the area enclosed by the luminosity curve .
Hence the area of the luminosity curve represents the totalbrighbnessofthelightwhichisformedintothespectrumandis , therefore , aconstanbwhate.vertheconditionofthevisionofthepersonwhomakestheobservationsThisisatonceapparentifwerememberthatifthe brightness of the whole recombined spectrum is compared with the comparison white , since these whites are derived from the same source and must , therefore , have exactly the same composition , a setting which appears correct to one person must also appear correct to any other , whatever the differences which may exist between their vision .
The fact thaG with a given spectroscope and source of , if the scale .
used for measuring the white is the same for all measurements , the area of the luminosity curve will be always the same is of great utility , since it allows us to correct for any variation in the unit in terms of which the white is measured , such as may occur owing to the mirrors , lenses , etc. , becoming tarnished .
It may be mentioned here that when a colour-blind observer makes a luminosity-curve determination I have often obtained my own curve at the same time .
The two sets of measurements being made immediately one after the other there is no time for the instrument to alter , and thaareas of the curves obtained have always been the same , although the shapes of the curves are often entirely different .
S 2 .
Calculation of the Form of the Curve for Cotour-Deficient Observers.\mdash ; Before giving the luminosity curves obtained in the case of colour- deficient observers , it will be convenie1lt to calculate what would be the for1n of the curves on certain assumptions .
We shall then be in position to judge of the accuracy of these assumptions by comparing the calculated curves with those obtained by observation .
Starting from the well-established fact that all colours can be matched by a combination , in suitable proportions , of three properly selected colours , Thomas Young and Helmholtz developed a theory of colour vision which supposes that the machinery by which oolour is perceived consists of three sets of mechanism , and that the sensation produced by of any .
is the sum of due to the individual mechanisms .
Or , as we may say , the sensation produced is the sum of the separate sensations produced by these separate mechanisms .
It is hardly necessary here to consider the line of reasoning by which the amount of each sensation produced by the different colours of the spectrum can be derived from experiments in which a coloured light of one wavefor the electric arc Area of red sensation curve 579 Area of green sensation curve 248 Area of blue sensation curve 'Phil .
Trans 1906 , , p. 333 .
having and Abnormal Colour Vision .
409 We have now to consider the manner in which the sensation curves of a person who has defective colour vision differ from the normal .
In several papers , Abney has forward evidence that , at any rate , in the case of the more dinary forms of colour defect , either the red sensation curve or the green sensation curve is modified , and that this modification is such that the shape of the curve remains unaltered , each ordinate being reduced in the same proportion .
* Thus , if the red sensation is deficient , and , say , the ordinate of the red sensation curve for any given wave-length is half the corresponding ordinate for the normal , then for all other wave- lengths the ordinates of the red se1lsation curve are also half those of the normal red sensation curve .
Hence it follows that the area of the red : sensation curve musb also be half the area of the corresponding curve for the normal .
If we possessed any method of measuring in absolut , units the sensation produced when light of a given intensity enters the eye , we should be ] to tackle the question as to whether the maximum ordinate of the sensation : curves was the same for all persons .
A certain amount of evidence on this matter has been obtained by Abney by the study of the minimum intensity of light required for vision .
For the present , however , we are confining ourselves to a study of the relative sensations produced by given amounts of light of dtfferent colours for each observer separately .
Thus we take some one kind of light as a standard and compare the relative stimulation of the sensations produced by other kinds of light with that produced by this standard light .
This is what is done when detsrmining a luminosity curve , the particular light chosen as a standard being white light , i.e. the light which we are going to use to produce the spectrum which supplies the different colours which are to be used iu the measurements .
Having to use white light as a standard , when we come to compare the measures made by a person having defective colour vision with those made by a normal , the matter is complicated by the fact that the unit with which the comparison is made , that is the'white , is pot the same for both .
The defective sensation will not only affect the sensation received from the coloured light , but also that received from the Abney 's theory as to the way the sensation curves of persons having defective colour vision differ * Another way in which we might have a departure from the normal would be if , say , the green sensation curve were displaced in such a way that its maximum occurred at a different wave-length .
There seems evidence that such a displacement may occasionally be met with , and the author hopes shortly to discuss this question in another communica- tion .
He has calculated the luminosity curves corresponding to such displacement and finds that they all intersect at neighbouring points , which do not coincide with the point referred to on p. 411 .
the wave-length is , and that and are the ordinates of the normal ] red and green sensation curves for this colour , the ordinates q for the observer will be and .
The total sensation produced by the !
colour will be the sum of the two sensations , that is , for the normal it will be and for the observer The sensation produced by the comparison white in the luminosity measurement will be proportional to the sum of the areas of the red and green sensation curves .
Hence , if we represent the areas of these curves for the normal by and respectively , the sensations produced by the white for the normal will be , and for the colour-deficient will be Thus the brightness of the coloured light is for the .
observer reduced in the ratio ) , while that of the white is reduced in the ratio .
Let be the intensity of the white when the normal observer makes the luminosity setting and the intensity of the white when the colour-deficient observer makes the setting .
Then we have .
( 1 ) and , ( 2 ) where is a constant which depends on the unit used to measure the intensity of the white light .
the case of a normal eye the total luminosity due to the blue sensation when white light enters the eye is only about 1/ 200 of that due to the red sensation .
results together with the normal luminosity curve are shown in fig. 2 .
It will be observed that all the luluinosity curves intersect at one point , , which corresponds to S.S.N. or a wave-length .
The i- condition that the luminosity should be the same for the normal as for , say , the .
observer is that [ equations ( 1 ) and ( 2 ) ] or or , ( 4 ) that is at a wave-length such that for both observers the ratio of the ordinates of their red and green sensation curves is the same as the ratio of ersons [ orresp nntersection independent o amount or kind of the deficiency in colour sensation of the observer .
FIG. 2 .
The curves given in fig. 2 depend on\mdash ; ( 1 ) Ths accuracy of Abney 's sensation curves .
( 2 ) The correctness of Abn ey 's theory that in the case of the ordinary types of total or partial red or Jreen colour-blindness the ordinates of one of the sensation curves are all reduced in the same proportion , and ( 3 ) The additive property which has been assumed and which involves the corollary that the areas of the luminosity curves obtained by normal and colour-deficient persons are the same .
Thus if it can be shown that the observed luminosity curves of persons who are colour deficient agree with the calculated curves it is strong evidence in favour of the correctness of the above three assumptions .
To test this point the observed luminosities obtained by colour-deficient observers will be shown plotted on diagrams containing the calculated curves in fig. 2 , so that it will be possible to see at a glance whether ( 1 ) the observed points all lie on one of these curves or lie uniformly between two adjacent curves indicating that the observer belongs to a class intermediate The wave-length at which all the curves will intersect depends on the distribution of light in the spectrum employed in the experiments , i.e. on the source of light .
them were published* before the sensation curves used in the calculations had been obtained .
It is also open to anyone to check the results by reference to the original papers .
[ Secondly , use is made of a selection from a large number of fresh measurements obtained in a different manner .
Since the equality of brightness method of obtaining the luminosity curves described above , involving as it .
does the comparison of the brightness of two lights which differ greatly in colour , is a process which requires considerable practice , although given such practice Abney 's numbers show that concordant measures can be obtained , : the flicker method originally described by Ferry , has been utilised for the : determination of luminosity curves .
In one arrangement which has been found to give good results , even with quite inexperienced observers , cylinder of diameter rotating about a horizontal axis is placed between the movable slit and the screen on which the light is received .
Segments are cub out of the circumference of the cylinder so that the screen is alternately illuminated by the coloured light and the comparison white , both kinds of light falling on the same part of the screen .
In order to obtain satisfactory measurements it is essential that the alternations from colour to * Sir W. Abney has stated in one of his papers that the numbers given in the papers which are quoted were all obtained before he had determined the sensation curves .
'Amer .
Journ. Sci 1892 , vol. 44 , p. 192 .
VOL. LXXXVIIL\mdash ; A. 2 FIG. 3 .
'Roy .
Soc. Proc 1910 , , vol. 83 , p. 472 .
In figs. 3 and 4 we have examples of persons who have R.S. and : G.S. respectively .
In fig. 6 observer has G.S. and observer has R.S. .
In figs. 6 and 7 we have two examples of persons having R.S. and one having G.S. It will be noticed that when the observed points are plotted on such a scale that the area of the curve drawn through them is the same as the *All the numbers given in this paper were obtained when the observer was dark adapted .
On plotting the numbers iven in the ' Philosophical Transactions ' for V. H. , it was obvious that some error had crept in , and on submitting the matter to Sir W. Abney , he said he had already moticed that , owing to a mistake when reducing the scale readings of the instrument to S.S.N. , the numbers as given were erroneous .
He has kindly looked up the original asurements and the points given in fig. 6 are obtained from these numbers .
FIG. 6 .
Roy .
Soc. Proc 1910 , , p. 460 .
, ' Roy .
Soc. Proc 1910 , , p. 472 .
Abnormal Cobur Vision .
417 FIG. 6 . .
B. , ' PhiL Trans 1892 , p. 556 .
V. H. , ' Phil. Trans 1892 , p. 555 .
FIG. 7 .
, ' Boy .
Soc. Proc 1910 , , vol. 83 , p. 466 .
FIG. 8 .
No. 2 .
curves were obtained with an arc in which the positive carbon was horizontal , so that the crater is directly facing the slit of the spectroscope .
The equality of brightness curves were obtained using an arc with nearly vertical carbons , so that the crater was turned more or less edgeways towards the slit .
Most of the difference between the two curves is , however , a real difference , due to the difference of method .
Tbis difference is most marked in the green and May 6 , 1913.\mdash ; This is the mean obtained by 10 observers .
* , foveal .
No. 6 , parafoveaL rightness oparts opectrum omparison white varies tntensity ooured light Hence , oted tarying tintensiGy oxamined bseries o the comparisons with white were made , while , in my experiments , the brightness of the coloured light , and hence also of the comparison white , was very much less at the ends of the spectrum than at the middle .
$ In fig. 8 , Observer 1 has about , while Observer 2 has about In fig. 9 , Observer 4 has about , and Observer 3 , .
In fig. 10 , the points marked by ashow that Observer 5 has a little over Observer 1 has a shortened spectrum , a neutral point , and he mistakes red for green , and vice .
He matches a mixture of green and violet light with white , and any amount of deep red could be added to the mixture without his being aware that any change had been made , although this addition of red caused the mixture , to a normal eye , to change from a bluishgreen to a bright red .
He is quite unable to correctly distinguish the colours of ships ' lights .
Observer 2 has a neutral point , and calls the green of this part of the spectrum white .
He often calls a green light white or a white light green , and , with lights of small intensity and small angular magnitude , he occasionally confuses green and red .
Observer 3 frequently calls white green and green white , and , with lights of small angular magnitude , he occasionally calls red green .
Observer 4 has a shortened spectrum and confuses red and green .
With ships ' lights , at distances up to 2 sea miles , he made over 20 per cent. of mistakes .
Observer 5 has a neutral point and calls green white and white green .
If a little white light is added to any green , he calls the mixture white , although to a normal eye it still appears a fairly saturated green .
S 5 .
In the preceding two sections examples have been given of what may be termed " " normal\ldquo ; luminosity curves for persons having both normal and defective colour vision .
The great majority of the luminosity curves obtained belong to this class .
In a few , cases , however , both of persons possessing normal colour vision and of those possessing defective colour vision , the luminosity curves differ in that they do not agree with any of the curves given in fig. 2 , nor do they lie uniformly between any adjacent pair .
Taking first the case of abnormal curves given by persons who have normal colour vision .
In and 12 , the points marked by , and through 'Phil .
Mag July , September , and December , 1912 .
through toint Purve passing cbove toint ahements mbservers Ieither curves passwhich tdotted 1correspond tosity m ' * FIG. 11 .
, foveal .
No. 6 , parafoveal .
FIO .
12 .
, foveal .
No. 7 , parafoveaL border subtended an angle of at the eye of the observer , while the external border subtended an angle of .
A small white dot at the centre served to fixate the eye .
Although this arrangement is more difficult to use than the small square , yet with a little practice fairly good .
settings could be obtained .
The points marked in figs. 11 , 12 , and 14 give the results obtained for the parafoveal region for Observers 6 , 7 , and the author .
The first thing that strikes one is that in the case of Observer 7 the luminosity curve is practically the same for the parafoveal and the foveal regions , and that both of these agree with the author 's parafoveal curve .
Next , in the case of Observer 6 , the parafoveal curve more nearly agrees with the foveal curve for the normal .
This distinction is brought out very clearly in fig. 15 , where the ratio of the parafoveal luminosity to the foveal for No. 7 and W. W. are shown .
It will be seen that for Observer 7 the rati is practically constant and equal to unity throughout the spectrum , .
while for W. W. the ratio increases rapidly as the blue end of the spectrum is reached .
It would thus appear that Observer 7 has no more pigment at 'Phil .
Trams 1860 , vol. 160 , p. 76 .
S. S. N. FIG. 16 .
1o Observer 6 cannot make reliable settings beyond the blue-green , the light 3 appearing to him too feeble .
requireslittleornoviolet , eason being twing tacular absorptionmatching awhite beans omixture ogreen , iolet light hObserver 6appears tuite normal colour 8ense , although w the white is to him much\ldquo ; yellower\ldquo ; than to a normal eye , and hence the ; mixture of red and green which to the normal looks yellower than ; S white appears to him a correct match .
He is quite good at matching blues ] and violets .
Observer 7 has also normal colour sense , although for his white match he uses more violet than the normal .
In figs. 16 and 17 are shorn the flicker luminosity curves of two observers who have defective colour vision , the green sensation being in each case in defect , and who , in addition , appear to have more than the normal amount of pigmentation of the fovea .
In fig. 18 Observer has more pigmentation than the normal , and his G.S. is practically zero .
Observer N. W. , on the hand , is defective as to red sensation but has less pigmentation than the normal .
It would be very interesting to examine the parafoveal luminosity curves for these colour-deficient observers , but unfortunately up to now it has been impossible to get them to devote the necessary time to make the observations .
The parafoveal curve for Observer 5 is indicated by the points marked in fig. 10 .
The pigmentation of this observer appears to be normal .
In fig. 19 are given the luminosity curves of two persons who were * The values given in figs. 18 , 19 , 20 were all obtained by the lity of brightness method .
A displacement of the green sensation curve towards the red end of the spectrum would give a luminosity curve very like those ined by Observers 8 and 9 .
A study of their parafoveal luminosity curves might enable one to say whether they are cases of green deficiency plus abnormal macular pigmentation or cases oi such dkplacement .
: ; having and Abn : S. ; FIG. 16 .
, foveaL FIG. 17 .
, foveal .
suffering from tobacco scotoma , and it will be observed that the curves are of an entirely different type to any of those previously given .
In fig. 20 FIG. 19 .
, ' Roy .
Soc. Proc 1891 , vol. 49 , p. 502 .
C. , ' Roy .
Soc. Proc 1891 , vol. 49 , p. 493 .
having and are given the curves for two observers due to disease ( atrophy of the optic ner very nearly rees with the normal , A resembles that of Observer 7 .
It disease Ghe colour-blindness which is to that which is found in cases of case comparison of the luminosity curv seems as if it might give us valuaOle in perceiving apparatus .
9 .
' : Sg Miss W. , ' Roy .
, vol. 49 , p. 507 .
W. S. , ' Boy .
Soc. Proc 1891 , , p. 498 .
The examples of the luminosity rVeS of different colour-deficient persons which have been given form only a small proportion of the cases which the author has examined .
They will , however , be sufficient to show that the : assumptions on which the calculated curves are based must , in the main , be true .
Abney 's theory that the different classes of the more commonly occurring forms of congenital colour defect are due to different degrees of deficiency in the red or green sensations is so strongly supported as to practically amount to a complete proof .
It must be remembered that the *The eye tested had become " " colour blind\ldquo ; after a stroke of paralysis ; the other eye was still normal .
VOL. LXXXVIIL\mdash ; A. 2 428 Prof. W. H. Brag and Mr. W. L. Brag .
[ Apr. investigation given above has been confined to that portion of the on the red side of the blue .
If we wish to extend our investigation into blue and violet then we should have to take account of the effect of the blue sensation .
Further , although it seems quite clear that the major part of luminosity perceived in that part of the spectrum including the red , yellow and green , at any rate with fairly bright spectra , is due to the red and greeq sensations , there is much evidence to indicate that , at any rate in the green and blue , there is a fourth sensation which may be called the fundamental : .
white sensation .
This probably corresponds to the rod-effect of von Kries , which would have to be taken into account if extremely minute differences of luminosity were under consideration , or were with spectra of .
small intensity .
The Reflection of -rays by Crystals .
By H. BRAG , , F.B.S. , Cavendish Professor of Physics in the University of Leds ; and W. L. BBAGG , B.A. , Trinity College , Cambridge .
( Received Apri17 , \mdash ; Read April 17 , 1913 .
) In a discussion of the Laue photographs it has been shown*that they may conveniently be interpreted as due to the reflection of -rays in such planes within the crystal as are rich in atoms .
This leads at once to the attempt to use cleavage planes as mirrors , and it has been found that mica gives a reflected pencil from its cleavage plane strong enough to make a visible impression on a photographic plate in a few minutes ' exposure .
It has also been observed that the reflected pencil can be detected by the ionisation method .
For the purpose of examining more closely the reflection of -rays in this manner we have used an apparatus resembling a spectrometer in form , an ionisation chamber taking the place of the telescope .
The collimator is replaced by a lead block pierced by a hole which can be stopped down to slits of various widths .
The revolving table in the centre carries the crystal .
The ionisation chamber is tubular , 15 cm .
long and 5 cm .
in diameter .
Tt can be rotated about the axis of the instrument , to which own axis is perpendicular .
It is filled with sulphur dioxide in order to increase the ionisation current : both air and methyl iodide have also been used occasionally to make sure that no special characteristics of the gas W. L. Brag , ' Proc. Camb .
Phil. Soc vol. 17 , Part I , p. 43 .
W. H. Brag , ' Nature , ' Jan. 23 , 1913 .

