120 On the Change in Hue of Spectrum Colours by Dilution with White Light , By Sir W. de W. Abney , D.Sc .
, D.C.L. , F.R.S. ( Received October 15 , \#151 ; Read November 25 , 1909 .
) That a change takes place in the hue of the spectrum colours when white is added to them has been noticed , but I am not aware that any measurement of the extent of the change has been made .
The need of a quantitative investigation has been forced upon me by the fact that in a research on the colour of fluorescence in various bodies the value of the dominant colours obtained by a photographic method did not correspond exactly with that obtained from visual observations .
The photographic results also showed that mixed with the dominant colour of the fluorescence was more or less white light , and it occurred to me that this added white might account for the difference .
The colour-measuring apparatus , which has been described in various papers and most recently in the paper " On a Modified Apparatus for the Measurement of Colour/ '* was employed in the investigation , a slight modification in the method of obtaining the two similar spectra being carried out .
In the apparatus described in that paper a portion of the beam of light which had traversed the collimator , the prisms , and the lens which brought the spectrum to a focus was diverted by means of a partially silvered mirror placed in the path of the beam .
In that used in the present instance only the lower half of the beam was deflected at right angles to the axis of the beam , and again deflected nearly parallel to its original direction .
Both deflections were made by silvered mirrors .
The two spectra were of approximately the same intensity .
Two colour patches could now be formed side by side on a white surface when slits were inserted in each spectrum .
The beam of white light reflected from the first surface of the first prism could be thrown on either of the patches .
In the investigations here given the white was thrown on the right-hand patch which was produced by the spectrum formed from the diverted beam .
The two colour patches overlapped each other , but the two coloured surfaces were caused to touch each other by inserting a rod in the path of the beams .
Another rod also cast a shadow from the white light , so that the left-hand patch was free from any mixture of white .
Both spectra were accurately scaled , so that the wave-lengths in both were * ' Phil. Trans./ A , 1905 .
Change in Hue of Spectrum Colours , etc. known , and the same colour could be placed in each patch .
[ The patches of light were about 1J inches square .
] A qualitative examination of the patches to which white light had been added was first undertaken .
The luminosity of the white light was made of about half the luminosity of the D light of the diverted spectrum , so that the red and parts of the green had , of course , a larger percentage of white added to them than had the yellow , the yellow-green , and the orange .
Two patches of the same red were matched in intensity , and to the right-hand one white was gradually added .
It was seen that the hue was changed , and that the mixed light was certainly yellower than the original colour .
When the patches were orange the colour became decidedly yellow , and this change in hue continued until Scale No. 48*7 ( A , 5772 ) was reached , when no change in hue could be noticed .
Passing the slits into the yellow-green , the colour lost much of its green , and when full green was under examination the green became a yellow-green .
All through the green part of the spectrum this yellowing was apparent , and in the blue-green , as far as Scale No. 36 ( A 5085 ) , the colour appeared to shift in hue towards the yellow .
In the true blue-green , about Scale No. 31 ( A 4886 ) , the addition of white seemed to make no difference in hue , simply making it appear rather paler .
At Scale No. 28 { A 4776 ) the mixture of colour and white made the blue become redder .
In the violet , the addition of white caused the colour to become bluer .
These changes were interesting , as throwing light on several discrepancies which have been observed in colour descriptions .
It seemed possible that the change in hue from near the red to the blue-green might possibly be measured .
Practically this would include by far the most luminous portion of the spectrum .
A patch of green light at about Scale No. 40 ( A 5270 ) was first examined , and different percentages of white wTere added to it .
With the addition of 50 percent , of the luminosity of the white ( D being 100 ) , it was found that an exact match in hue could be obtained by altering the colour coming through the slit of the other spectrum , and that a match in luminosity could be obtained by altering the width of the slit .
The match made indicated a hue approaching the yellow .
As less and less white was added the match gradually approached the scale number of the undiluted colour .
A series of colours between Scale Nos. 59*5 ( A 6780 ) and 36 ( A 5085 ) were examined , with the result which is shown in the annexed table .
It now remained to ascertain if there was any law which could be applied to give the changes in hue .
The first point that was evident was that the Scale No. 48*7 ( A 5772 ) had something to say to any law .
At this point 122 Sir W. de W. Abney .
Change in Hue of [ Oct. 15 , .
it is to be remembered the addition of white made no alteration in the hue of the colour .
On examining the table in the paper above referred to , it was at once seen that at that point of the spectrum the proportions of red to green were exactly those of the proportions .
existing in white light .
This seemed to give a clue to the change in .hue that takes place .
[ It may be remarked that Scale No. 48*7 is the place where the red and green curves of equal stimulation of the three sensations cut one another .
] It seemed probable that the change in hue in the region of the spectrum under investigation might be due to the addition of the red and green contained in the white , the blue being a negligible factor .
To make this hypothesis and its results plain , parts of the table ( one or two small corrections , due to slips in calculation , being made in certain figures ) are reproduced .
Columns I , II , and III are the prismatic scale numbers , the wave-length and the luminosity of the electric arc ( crater ) spectrum respectively ; IV , V , and VI give the percentage composition of the colours in terms of the red ( R.S. ) , green ( G.S. ) , and blue ( B.S. ) sensations %r VII , VIII , and IX are the luminosities of B.S. , G.S. , and B.S. ( derived by multiplying IV , V and VI by Column III and dividing by 100 ) ; X and XI are VIII and IX multiplied by 2*3 and 178 respectively , to give the curves of equal stimulation to form white ; XII , XIII , XIV , and XV are the three sensations existing in the colours , together with the white ( XV ) inherent to them ; XVI , XVII , and XVIII are the percentages of the sensations existing in the colours , and XIX is the white which belongs to these different percentages .
In making tests as to the value of the hypothesis , the proportion of B.S. to G.S. in white was taken as 69 to 31 .
In one set of observations the value of the white used was 0*6 the luminosity of D , which gave the proportion of 41-4 B.S. to 18*6 G.S. A rotating section was used to get other percentages of white .
The following is a specimen of the calculations made for one colour with different mixtures of white :\#151 ; Scale No. 46*23 ( A 5611 ) .
B.S. G.S. The proportion of R.S. to G.S. in luminosities in scale No. 46-23 is ... ... ... ... ... ... ... ... ... ... ... ... 55-45 + 32-60 Addition of 60 W. ... ... ... ... ... ... ... ... ... ... ... .
4140 + 18 60 Total E.S. and G.S. is ... ... ... ... ... ... ... ... ... ... 96-85 + 50-20 Converting this total into percentages , we get Scale No. 46-23 + 60 W ... ... ... ... ... ... ... ... ... = 65'44 +34-55 .
Table I. 1909 .
] Spectrum Colours by Dilution with White Light .
123 XX .
( 'N'S'S ) .Jaqranu 0jbos p.repm ?
ig rH M O \#171 ; 0*X1 so 00 #\gt ; o ^ o lC iO iQ 00^^010 Tt Tt\lt ; T*1 Tt\lt ; 00 CO H ( M O co co co co co GO SO rH M O \lt ; M\lt ; M\lt ; M\lt ; M\lt ; M coto^^o H rH rH H H GO SO rH M O xix . !
1 IIIAX 'ITAX 'IAX ui saSBjuao-idd oumiorjippT ?
0'liqM I 1 I II 1 1 1 8-2 34 *6 70 150 135 191 *4 199*2 173*5 155 *8 150-8 CO kO O 00 CO SO SO \#169 ; \#169 ; rH rH M SO rH S3- 111 II 1 1 1 IIIAX u Hi M \#163 ; * B.S. 1 1 1 1 1 1 1 1 INI !
|I ?
S ?
O H CO so 12 -63 31 -6 89*5 85 30 -3 28 -2 26-9 28 28 28 GO GO GO 00 GO M M M M M XVII .
G.S. 1 1 1 l 4-5 9'5 15*8 25 32-9 40*1 48-9 59-9 72 -6 89-1 99 -6 98-31 96 -24 93-18 87 -37 68-8 10 -5 II 1 1 1 1 1 1 1 1 XVI .
3 o o o o o o p \#187 ; c m Oi iC o ^ \gt ; o 05 05 05 00 t- 05 rH r\#151 ; 1 N05HON .o kO IQ H W 05 \#169 ; 1 1 1 1 15 69-7 \lt ; X\gt ; iH rH CO M M M t- t- N- M M M M M trNNNtr X Luminosity of colour sensations together with the white .
\#166 ; s ' 1 1 1 1 1 1 1 1 kO rH 05 kQ 00 N CO O H 00 M CO CO M kO CM kO rH SO 05 kO CO CM rH M M r-r O O 0-15 0-03 l l l l l XIV .
B.S. I 1 1 1 1 1 1 1 1 II 1 I 0*031 0*088 0-125 0-155 \#171 ; NyKfKN o o o o o 0-201 0*180 0 *154 0-126 0-098 0 -073 0 *051 0-039 0-028 0-022 XIII .
G.S. I 1 1 \#171 ; m p p O WNiOlO rH CM p N- \#169 ; M TtH 05 lO rH 00 kO WWWhh 1\#151 ; \lt ; rH rH rH CM \#169 ; HOOkCCOW 1-33 0 53 0-03 1 1 1 1 1 1 1 1 1 1 XII .
i CM OWN 1 20*79 ' 47*75 72 -40 80-64 75 CM ip 00 00 skkb'* *1111 111 ? ?
o o rH CO 1\#151 ; 1 05 05 CO kO p rH CO 00 CM O O O O O ; 0-187 : 0 *130 0-101 0-072 0-057 R *811 x *s*a 1 1 1 1 1 1 1 1 5-18 16 21 -03 21 *80 20 16-14 18 22 25*74 30*12 SO kO CO SO kO 05 M SO 05 SO kO CO SO CO H CO rH rH rH rH 05 rH H CO rH rH \#169 ; rH rH rH SO\lt ; Mtr(MI\gt ; CO 00 M M tH 12*49 9*08 6-91 5-00 3-92 M *$.\amp ; \gt ; ' *S*D 1 1 1 00 00 00 rH iH rH CO kO O^ONlON r-t CO kO H SO CM O H M \lt ; X\gt ; \lt ; X\gt ; -p CO iO O CO kO t- N- SO uO 42-30 29-51 18 -08 10-97 7-08 CO rH M 00 CO O0\#174 ; WHW rH M rH O O M 00 05 M 99 \ \ \ \#169 ; o 1 1 1 1 1 a Luminosity of sensations .
B.S. 1 1 1 1 I 1 1 1 0*029 0-090 0-118 0*122 0*112 0-091 0-101 0*124 0-145 0-174 0*202 0-243 0-262 0*247 0*234 0*202 0-180 0-154 0-126 0-098 0*073 0 -051 0-039 0-028 0*022 VIII .
\#171 ; 5 o 1 1 i rH lO \#169 ; SO \lt ; M \lt ; M \#169 ; CO o WNJO^J 00 O O kO 05 l\gt ; N CO N O H W H N H CO CO CO M CM SSglD\amp ; g 2-03 1-15 0*53 0-27 o-io Tt* rH SS 1 1 1 1 II 1 1 VII .
3 ip ON 20*79 47*75 72-40 80 *64 75 SO O 1H o rH \lt ; M kO SO 00 kO " H CO rH kO SO kO rH CO M pppSp N H SO CO W 1-76 1-41 1-15 0-91 0-77 0-62 | 0 -51 !
0-392 | 0 -334 0 -253 N o iH SO tr .
QOMONkO tH tH tH O O o \#169 ; o o o p Percentage composition of colours in terms of sensations .
B.S. 1 J 1 I !
1 1 I O ^ 00 CO H CO O iO 05 !
M O H iH i\#151 ; 1 \lt ; M O O O O O 0*252 0-420 0- 870 1- 700 3*160 O 00 rH rH 00 \lt ; M \#169 ; rH SO CM kOOOCONH rH iH M M SO rH kp t\#151 ; rH CO kO GO GO \lt ; M M M M M CO 00 00 GO 00 ( M M M M M G.S. 1 1 1 kp ko oo h rH 05 kO kO 05 p tH M ( MNnrjlOO CO CO rH rH rH HCOO HlOWHCD H CO lO CO ^ kO kO lO kO kO p p 00 O rH GO SO 00 kO rH M iH 9 9 , , rH M O 1 1 1 1 1 1 1 \#163 ; 3 \#169 ; o o o o o iH\lt ; rH rH p p 03 \#174 ; kO O H KJ 05 05 05 GO 1\#151 ; CM t- H CO H H kO N W 00 kO H so SO kO kO iO kO 00 05 t\#151 ; rH \#174 ; ONHW GO SO CO CM CM rH rH rH rH rH SO M SO CO M WOippN rH O 00 kO O rH kO kfO SO N- GO 0 ' rH M M M M N NN N N \lt ; M M M M M t " .
tr .
tr .
a .imu^ -oads jo ^^isouxuin'-j ip O M N- M O 00 \#169 ; \#169 ; p NNiONO 05 00 t~- SO kO 36 24 14*2 8-5 5-7 kO 00 \lt ; 05 rH iH rH M rH rH iH SO O SO kO kO GO tr .
lO rH CO O O O \#169 ; O SO 00 rH o SO M rH tH r\#151 ; i O o o o o o s \lt ; 7217 6957 6728 6521 6330 6152 5996 5850 5720 5596 5481 5373 5270 5172 5085 5002 4924 4848 4776 4707 4639 4578 4517 4459 4404 4349 4296 4245 4198 4151 4106 4062 4010 M ( 'N'S'S ) 'Joquinu 0JBOS pj'epU'B'Jg rH M \#169 ; SO SC SO OD T ^ ( M O iO lO vC 00*^^(NO ^ ^ Tfl 00 ST rH M O CO CO CO CO CO M M M M M GO SO rH M O rH iH rH rH rH GO SC rH M O 124 Sir W. de W. Abney .
Change in Hue of [ Oct. 15 , Turning to Table I , Columns IV and Y , we find that this would give a colour Scale No. 47'3 .
It will be seen that this is the same colour that matched the mixture visually .
Taking half the white ( he .
30 white ) , we get by the same method of calculation a scale number whose percentage is 64:5 + 35'5 .
This gives a colour whose scale number is 46'92 .
It will be seen that the colour matched 46'89 .
Again , using one-quarter white ( he .
15 white ) , we get a percentage equation\#151 ; E.S. G.S. 63-1 + 36-9 which is the proportion of K.S. to G.S. in Scale No. 46-33 ( X 5618 ) .
The match found was 46-3 .
For all the other scale numbers mixed with different proportions of white the calculations were made in the same way , and the table attached shows the results of the matches and calculations .
Table II.\#151 ; White 60 .
60 W. 30 W. 15 W. 7*5 W. 3-75 W. 1-87 W. Scale No. Calc .
Found .
Calc .
Found .
Calc .
Found .
Calc .
Found .
Calc .
Found .
Calc .
Found .
56 53 -6 53 -9 54 -6 54 -3 55 *2 54-9 54 *45 \#151 ; \#151 ; 52 -19 52 -6 53 53 -15 54 54 50 *6 \#151 ; \#151 ; 50 -13 50-12 52 -9 \#151 ; \#151 ; 51 *7 51 -6 52 -3 52 -2 52 -7 52 *7 46 *23 47 *3 47 *3 46 -92 46 -89 46-33 46 -3 43 *65 46*17 46 -3 45 45 44-6 , 44 -68 44 -14 44 -05 41 *1 \#151 ; \#166 ; \#151 ; 44 -5 44 -75 43 -2 43 42 42 *3 41 -7 41 *43 38 *5 \#151 ; \#151 ; \#151 ; \#151 ; 41 -6 41 -6 40 -3 40 *4 39 *5 39 *3 35 -92 \#151 ; \#151 ; \#151 ; \#151 ; \#151 ; \#151 ; 40 -1 40 -2 38 -5 38 -57 37-4 37-3 Another set of matches and calculated values will be found in Table III .
Table III .
43 W. 21-5 W. 10 75 W. 5*37 W. 2-68 W. Scale No W. No. Calc .
Found .
Calc .
Found .
Calc .
Found .
Calc .
Found .
Calc .
Found .
57 -47 57 *47 55 *96 55 -06 55 -92 56 -49 55 1 55 -1 51 -76 51 -82 52 -8 52 -8 53 -1 53 -5 54-1 54 *2 54 '47 54-9 49 *7 49 -7 49 -34 49 *34 49 *6 49 -5 47 47 47*58 47 *6 47 *4 47 -4 44 -83 44-3 44-3 46 *2 46-16 45 -6 45 -4 45 -14 45 -16 44 -9 41 -5 41 -5 45 *36 45 *36 44 -1 44 43*1 43 42 -35 42 *45 38 -2 38-2 44 -8 44 *83 42 -9 42 *93 41 *27 41 *39 40 -37 40 -33 36 -1 36*1 45 -2 44 *8 43 *2 43 *2 41 41 -1 39 *3 39 *5 1909 .
] Spectrum Colours by Dilution with White Light .
125 The results are shown graphically in the figure .
[ In the above tables where there are no results given for the greater proportions of white ( as in Table I for Seale No. 56 ) , the match became uncertain owing to too great a proportion of white being added .
] The results here recorded bear out the hypothesis made .
They show that in the parts of the spectrum under measurement the value of the blue sensation is unimportant as regards hue , when matches of impure with pure colours have to be made .
It is also worthy of remark that there is a considerable range of spectrum colours which can , by adding different proportions of white , be caused to match a pure colour .
This points to using a spectrum in a room free from all white light , and with prisms as free from dirt as possible to prevent the white illumination of the surfaces .
Another useful fact is this , that if S.N. 487 ( \5772 ) be matched by a mixture of red ( say at the red lithium line ) and any green , the amount of green sensation in the green employed can be easily calculated , as the blue has no effect on the match of hue .
For although the colour to be matched may contain red as well as green sensations , yet the former is accounted for in the white ( see Table I , Columns XIII and XY ) .
It must be remembered that S.N. 48*7 is very readily fixed , from the fact that it is the colour which , with pure blue , can make a white to match the white of the crater of the electric light , quite regardless of the yellow spot difficulties .
In the determination of the equations to the three colour sensations , one of 126 Change in Hue of Spectrum Colours , etc. the first researches was to find the amount of inherent white in that colour which represented the colour of the green sensation mixed only with the white .
When that was found , the equation to make white in terms of the red , green , and blue sensations became an easy matter , and therefrom the amount of red and green sensations was easily calculated .
In some experiments made with the object of checking the amounts of red and green sensations in the colours ( see Table I ) lying between S.S.N. 48 ( \ 5720 ) and 36 ( X 5085 ) , the place of the colour to be matched was close to 48-7 and had 69'86 E.S. and 3014 G.S. This colour was isolated from one spectrum and matches made with the colours coming through two slits placed in the other spectrum .
One of these slits was placed at the position of the red lithium line and the other moved about in the green as required .
The matches were made by opening or closing the slits .
The following are specimens of the results . .
The red slit was placed at S.S.N. 59'8 ( X6501 ) , and the green slit at S.S.N. 40'8 .
The respective relative luminosities through equal slits were 81 and 55 ; the red slit had an aperture of 102 on an empyric scale and the green 27 .
The luminosities are therefore 102 x 81 = 857 and 27 x 55 = 148-5 .
The two luminosities added together = 234-2 .
As S.S.N , 48-7 contains 69-86 of R.S. , 234-2 has to be multiplied by this , the result being divided by 100 , which equals 163-6 of R.S. ( supposing , as shown before , that B.S. is negligible ) .
But there is 857 of R.S. from the red slit .
Therefore in the green colour S.S.N. ( 40-8 ) there can only be 77'9 R.S. , the remainder of the colour being 70"6 G.S. We therefore have the colour S.S.N. 40-8 represented by R.S. G.S. R.S. G.S. 77-9 + 70-6 or 52-5 + 47-5 .
Turning to the Table I , columns IV and V , we find that by other means S.S.N. 40"8 was found to contain 53'0 R.S.+ 47 G.S. Similarly , when the green slit was placed at S.S.N. 43-7 , which has a luminosity of 73-1 on the same scale and a width of 27 , and the red slit a width of 94 , by making similar calculations the colour ( 437 ) is represented from these observations by R.S. G.S. 57-7 + 42-3 , which is the same composition as that found from the table .
Again , with the green slit placed at S.S.N. 38 , the luminosity of which is 36 , a match was made when this slit had a width of 34 and the red slit of 102 .
The resulting calculations gave R.S. G.S. 49 + 51 Harmonic Tidal Constants , etc. In the table it is 48*65 E.S.+ 51*1 G.S. This indicates that the resulting equations obtained from these researches are trustworthy , and also opens out a method by which the sensation curves can be obtained with great facility .
The part of the spectrum from S.S.N. 64 ( A , 7220 ) to S.S.N. 48 ( A 4720 ) is readily obtained by ordinary methods , as is the portion from S.S.N. 36 ( A 5085 ) to the extreme violet .
The most difficult portion is from S.S.N. 48 ( A 4720 ) to S.S.N. 36 ( A 5085 ) , and this can be checked by the method indicated above .
Harmonic Tidal Constants for certain Chinese and New Zealand ' Ports .
By Thomas Wright , formerly of the Nautical Almanac Office .
( Communicated by Sir George H. Darwin , K.C.B. , F.R.S. Received October 19 , 1909 .
) Wei-hai-wei ( China ) , Woosung { China ) , Port Chalmers { New Zealand ) , Port Lyttelton { New Zealand ) , Wellington { New Zealand ) , and Auehland { New Zealand ) .
The following results have been obtained by the aid of sums placed at my service by the Government Grant Committee of the Royal Society .
Constants for ten other ports , obtained by similar grants , are given in vol. 71 , pp. 91\#151 ; 96 of the 'Roy .
Soc. Proc. ' The observations were reduced by the methods described by Sir George H. Darwin in vol. 48 , pp. 277\#151 ; 340 , and vol. 52 , pp. 345\#151 ; 389 .
The observations made at Woosung , Wellington , and Auckland wTere the times and heights of high and low water : those for the first port referred to the year 1902 ; those for the second to the year 1901 ; and those for the third to a period of twelve months , beginning on May 1 , 1900 .
The observations at Port Chalmers , Port Lyttelton , and Wei-hai-wei were hourly heights read off from automatic records .
Those for Port Chalmers referred to the year 1901 ; those for Port Lyttelton to a period of nine months , beginning on October 1 , 1901 ; and those for Wei-hai-wei to a period of seven months , beginning on July 8 , 1898 .
In dealing with observations of high and low water , the year 's observations were divided into four groups of about three months each .
Pour separate determinations of each tide were made , and a final evaluation obtained from the means of the four values of H sin k and H cos k. In the case of Woosung

