1902 .
] Contributions to the Study of Flicker .
313 " Contributions to the Study of Flicker .
Paper II .
" By T. C. Porter , M.A. , Eton College .
Communicated by Lord Rayleigh , F.R.S. Received May 14 , \#151 ; Read June 5 , 1902 .
In the first paper , * curves are given which show the rate at which a disc half white half black must be rotated in order that flicker may just vanish when the disc is illuminated by the different colours of a diffraction spectrum .
Also curves showing how the rate of rotation alters with the alteration in the angular width of the coloured sector when placed in the different colours of the same spectrum .
These last are reproduced with some additions and corrections in the present paper ( fig* 4)* It was shown in that paper ( a ) that as the stimulus applied to the retina increases in intensity , the impression produced retains its maximum value for a shorter and shorter time .
( 6 ) that a stimulus requires a finite time to produce its maximum effect , ( c ) that the duration of the impression on the retina undiminished decreases as the time during which the stimulus is applied increases , one of these quantities being roughly inversely proportional to the other .
There were also certain conclusions which confirm the work of others , and which it seems unnecessary to repeat .
Since this first paper was written , continued experiments have suggested fresh means and precautions for making the observations still more accurate , an these must be first described .
( 1 .
) The quantity of white light reflected from the black sector hitherto used was measured , by comparing its luminosity with that of a disc having two narrow white tapering projections , subtending known angles at the centre .
This was made to rotate , and viewed when flickerless , against a pitch dark background .
It was thus proved that the quantity of white light reflected by the Indian ink with which the black part of the disc was painted varied , when the disc was regarded at different angles , from 4 per cent , to 10 per cent , of that reflected by the white cardboard used , and although in the experiments already described , the disc had always been regarded from very nearly the same standpoint , when it reflected 7 per cent. , nevertheless it seemed better to use a black pigment which would diffuse what light it reflected more evenly .
The disc was therefore coated with a mixture of lampblack and alcohol , shellac varnish being added in just sufficient quantity to prevent the lampblack from rubbing off .
On careful measurement by the method already described , the quantity of white light reflected from this very ''dead black proved sensibly constant for a comparatively large variation in the angle of reflection , and equal to 4*1 per cent. It seems well at this point to state how the want of blackness in * 'Hoy .
Soc. Proc. , ' vol. 63 , p. 347 .
314 Mr. T. C. Porter .
[ May 14P the black sector affects the flicker curves ( such as those given in fig. 3 of the present paper ) , and perhaps the simplest way to make this plain , is to take a numerical example : suppose then that the amount of white light reflected from 360 ' of black is equal to that reflected from 24 ' of the white cardboard from which the disc is made .
then if the angular breadth of the white sector is , say , 60 ' , the total amount of white light reflected from the remaining 300 ' of black will be equal to the light reflected from 22 ' of white , and this light is distributed evenly throughout the black .
Now if this disc is made to rotate until it becomes flickerless , its luminosity or albedo will be rightly expressed as that of a mixture of 60 ' + 22 ' = 82 ' of white , and 278 ' of absolute black , and if a second disc were made with a white sector of 82 ' , and an absolutely black sector of 278 , this second disc , when flickerless , would appear of the same brightness as the other disc of 60 ' white , and 300 ' imperfect black , but\#151 ; and this is the point of the illustration\#151 ; the flicker does not just vanish at the same speed of rotation for the two discs , \#151 ; for flicker depends essentially upon contrast , and the contrast between the white and absolute black is greater than between the same white and imperfect black .
Let the whiteness or albedo of the white cardboard be expressed numerically by 100 , then the flicker in the case of the first disc is produced by the contrast between the 60 ' sector of white of albedo 100 , and the 300 ' sector of what is really dark grey , albedo 7'3 .
Thus 7'3 of the white sector 's brightness will contribute nothing to the flicker , for the disc may be regarded as being of the albedo 7'3 all over , there being added , where the white sector is , an additional albedo , not of 100 , but of 100 \#151 ; 7-3 = 92*7 .
In order , therefore , that flicker may vanish on a disc of 60 ' white , and 300 ' of absolute black , at the same rate of rotation as for a disc of 60 ' white and 300 ' of the imperfect black already mentioned , we must lessen the contrast between the white and black of the first disc by reducing the albedo of its white sector in the ratio of 92-7 : 100 , and this can be done experimentally by altering the distance of the rotating sector from the source of light .
Now the effect of altering the distance of the disc from the source of illumination is shown by the curves of fig. 4 , and since the greater the distance from the source of illumination the further the curve lies to the left in this figure , we see that the effect of the imperfection of blackness in the black sector of the rotating disc is to shift its curve further to the left than it would lie if the black were perfect , and thus , as is evident from the figure , to make it of a rather flatter form than it would otherwise be : it is also evident that unless the amount of light reflected by the black sector is large , the shift in the curve is small , and the difference in form very slight .
( 2 .
) The rate of rotation at which flicker appears just to vanish , depends , amongst other things , on the distance of the observer from 1902 .
] Contributions to the Study of Flicker .
315 the rotating disc ; the greater this is , the more difficult is it to see the flicker , hence in the experiments recounted in this paper , the disc was viewed at the same distance .
The distance chosen was that of most distinct vision , and though constant for each observer , differed slightly for different observers .
\#166 ; ( 3 .
) A black velvet mask was worn , both to avoid any possible complication produced by light reflected from the disc on to the observer 's face , and thence either into the observer 's eye obliquely , or back again on to the disc .
^4 .
) To determine the exact moment when all trace of flicker has vanished requires a considerable amount of practice : but is made far easier than it would otherwise be by the fact that flicker is much more easily seen by averted vision than by direct , so that if the eye be fixed upon the axis of the rotating disc , the flicker will be seen to disappear in the region immediately around the axis before it disappears from the parts of the disc near its circumference ; with practice this makes it possible to time the appearance of the flicker with very fair accuracy .
It seems a somewhat remarkable fact that although the perception of colour and form so rapidly deteriorate in passing from the fovea centralis towards the peripheral portions of the retina , the retinas sensitiveness to flicker should , on the contrary , increase .
Inasmuch as the perception of any motion is analogous to the perception of flickei , for in both cases the perception is caused by a change in the stimulus of certain retinal elements , it follows that the retina is increasingly sensitive to the motion of objects as their images on the retina lie nearer to its periphery .
May not this be due to the fact that it has always concerned mankind to notice motion before any other quality in matter , and especially motion of objects near the limit of his field of view 1 When once this motion has been perceived , the man turns his eye so that the image of the moving object is brought to the fovea , where its colour and form can be appreciatedit seems , therefore , a natural arrangement that the attention of an animal should , in the first instance , be arrested by the sensitiveness of the retina to the motion of an object which is not the object of direct vision at the moment its motion is perceived .
( 5 .
) In timing the disappearance of flicker , it is not only a good plan to fix the eye on the axis of the disc , as has already been said , but it is necessary to keep the gaze steadily fixed on the axis , for when flicker has just disappeared to direct and steady gazing , the smallest movement of the eye , either obliquely across , or round the disc in the direction of rotation will reveal the flicker momentarily , or for as long as the motion of the eye lasts .
The reason for the reappearance of the flicker scarcely requires explanation ; it is evident that any motion of the eye such as that mentioned above will retard the rate of Mr. T. C. Porter .
[ May 14 , passage of the image of the white and black sectors of the disc over the points of the retina , and will be equivalent , so far as the effect upon flicker is concerned , to making the disc rotate more slowly .
There are two kinds of motion the eye may make without causing the flicker to reappear\#151 ; the one radial , the other round the disc , in the opposite sense to that in which it is rotating .
Experiment verifies these conclusions , though in practice the radial movement is not very easv to accomplish .
y y A set of experiments was then made , paying due regard to all the foregoing precautions , to determine the exact number of revolutions of a disc half white , half black , at which flicker just vanishes , when the rotating disc is differently illuminated .
The experimental arrangements can best be understood from fig. 1 .
Fig. l. -p , - 1 L . .
S-t , :4 W5 " - j IS | , " In \lt ; S ' \#171 ; 1 e\#151 ; : S is the source of the illumination of the disc D : this illumination can be changed in two ways , viz. , either by altering the distance of D from S , or by changing the illuminating power of S itself .
The length of the bench was more than 5 metres , and the illumination of the disc varied from that given by a sperm candle burning 8*273 grammes , or 127*67 grains per hour ( which is meant when 1 candle-power is mentioned in this paper ) , at a distance from the disc of 4 metres , to the illumination given by an arc light of approximately 1600 ( \#177 ; 50 ) candle-power at a distance from the disc of ^ a metre .
lieturning to the figure , LL PP are inner and outer boxes , painted with dead black inside , having openings at 0 and O ' , with sharp edges , carefully blackened , wide enough to allow the whole of the illuminant to be seen from every part of the disc , when the latter was at its nearest to the illuminant .
The screens PP LL kept the light from S from falling on the walls and ceiling of the room , and care was taken 1902 .
] Contributions to the Study of Flicker .
317 that S was sufficiently raised above the working bench to prevent any rays of light from falling on the bench between S and D , and so getting reflected on to the disc or about the room .
By several screens such as SS ' the stray light was further confined , with so good a result that , when it was necessary to bow one of the standard forks used to determine the pitch of the note given by the twelve siren holes in the rim of the disc , it was done chiefly by , the forks being arranged in the order of their pitch on a neighbouring table and generally in charge of an assistant .
Q in the figure is a small screen with two eyeholes through which the observer looks at the disc , and which can be arranged at any desired distance from D ; some blackened screens , Fia .
2 .
$ 2-000 C determined with Sta.nda.rd Candle i c.p. L with Incandescent Lamp 25 c.p A with Arc Light / ,600 c.p. 4\gt ; 9$ S\amp ; 8 $ gg 8*3 !
8 , - , - . .
No. of rotations per second ( for disc % white , % black ) for to vanish .
CCC , receive any stray rays after they have passed D , so that the room may be described as very nearly dark ( when the air is not unusually dusty , for of course any dust in the path of the rays scatters some light ) .
The disc D is mounted on an electric motor , running smoothly , and since some of the sources of light used were electrical , it is important to state that the battery running the motor was altogether separate and independent of that which supplied the source of light .
It is also necessary to give some account of the means by which the powers of the illuminants employed were gauged , the range of illuminations being unusually great .
The photometer is shown in fig. 1 ; ABO and EDO are two right-angled prisms , having their faces BC and CD in one plane ; both these faces are finely and evenly ground : light from the two sources is incident normally on the 318 Mr. T. C. Porter .
[ May 14 faces AB , ED , and passing through the glass of the two prisms iS P totally reflected from their hypotenuses AC and EC , then falling oh H the ground-glass surfaces BC and CD .
The prism and the sources of light are so adjusted that the albedos of these faces are equal when the brightnesses of the two sources are as the squares of their distances 1 from C. First , then , for the lowest degrees of illumination , the sperm candle already mentioned was used .
This was next replaced by an incandescent electric lamp , of the common type , supplied by a mea-sured and constant current .
This gave , as a mean of five determine- I tions , the candle-power 25-06 .
A large coal-gas bat Wing burner was I then measured for candle-power against this incandescent lamp , and !
in turn , this gas-burner was balanced photometrically against the arc | iight which was used for the brightest illuminations , and which gave as the mean of three very careful determinations 1600 candle-power !
with a possible error of 50 candle-power , though the probable error did not in the writer 's estimation exceed 20 candle-power , if as much .
The gas-burner was used to connect the measurements of the arc and 25-06 candle-power lamp , because the arc and this lamp depended on the same battery of accumulators , and hence could not be compared directly .
The results of this part of the investigation are exhibited tabularly below , * and in fig. 3 by curves .
On the axis of Y distances are measured proportional to the logarithms of the different illuminations of the disc , half white half black , whilst on the axis of X can be read off the corresponding number of revolutions of the disc per second in order that ' flicker may just vanish , and it is at once evident that for all the illuminations between that caused by one candle at 2 metres and 12,800 times this illumination , the speed with which the disc must be driven in order that flicker may just vanish varies directly with the logarithm of the illumination of the disc .
The same logarithmic law seems to hold good for illuminations smaller than these , but the rate of variation is different though constant .
With regard to the apparently sudden break in the line , it may at once be said that it may in reality be only a rather sharp curve , but that there is an unexpectedly rapid change in the direction of the line the writer feels certain .
Each experiment was repeated at least three times for each observer ( of which there were never less than two , and ?
often as many as four ) , and , moreover , the whole set of experiments were repeated after an interval of 4 years .
# For this table , see next page .
1902 ] Contributions to the Study of Flicker .
319 Log. of illumination .
o-ooo 0- 248 0 602 1- 204 1 -806 1 -398 1- 617 2- 000 2-602 3-204 SI 8 8 8 N 00 ^ t- CO CO co ^ Th Number of revolutions of disc per second .
\#151 ; 17- 75 18 -08 18- 50 25-08 33-50 28 -00 32-00 35-50 42-66 50 -16 S 8 | 8 8 O \#187 ; c CD lO rH lO to *D CD t\gt ; Number of sound vibrations per second .
213 217 222 301 402 SSS2'S w CO rfi lO Q O rH i\gt ; O CD qo ic o CD CD X\gt ; QO f I Note given by 12 holes in the rim of the disc .
\#151 ; ;\#151 ; ~ 7~ . .
1 ; o .
\ * * *"gg * * * .
u .
* 1 1 1 ^ . .
: | g- * : :\#171 ; : || j| : : HI lo*o* oo .
H . .
\#163 ; * S * \#163 ; % % -\#174 ; \#163 ; *s n -1 ^ ^ ^ ^ ^ *\lt ; 1 H\#171 ; 0 fi ^ 11 1 1 J2 .* * !
1 .
I- . . .
* _ .
r-H + . .
r \ c8 * * * \#171 ; j j j j i \lt ; D \#174 ; \#169 ; .
\gt ; .
2 3 * - s=** Pr $ \lt ; 1 O ft D| above C5i2 , very little Mean between E and F above C512 F about 5 vibrations per second flat Q- a very little jjl A above C512 Degree of illumination of disc .
\ 9 V* 9 9 9 | - .
* s s 1 1 9 - ?
9 9 9 S S. 8 | | 1600-0 2844-0 6400-0 25600 0 51200-0 Distance of disc from source of light in metres .
4 3 2 1 0-5 4 3 2 1 0*5 ID T ?
CO \lt ; M r\#151 ; 1 O Source of light used .
A standard candle .
An incandescent electric lamp ( Robertson 's ) giving 25 candle-power .
Arc light of 1600 ( \#177 ; 50 ) candle-power .
IG .
3 .
Mr. T. C. Porter .
[ May 14 , * a O ^ 9 *\gt ; ^ '"2 ^ -o \#163 ; -\#171 ; o \lt ; i / , . !
' / 3lSS8SSSsss AnguLar magnitude of w. h Log w(360-w ) curves .
1902 .
] Contributions to the Study of Flicker .
321 Calling the illumination for which the direction of the straight line seems to change , the critical illumination , the equation to the upper part may be written dnfd log I = cot 39 ' ( approx. ) for the scale of the diagram , whilst below the critical illumination it is log I - cot if$2 ' 30 ' ( approx. ) .
It may be noticed that the angle made with the axis of X by the lower part , i.e. , 82 ' 30 ' , is a little more than twice the similar angle made by the upper part , i.e. , 39 ' .
Remembering that the scale of the logs ( to base 10 ) is that of one-tenth , that of the numbers expressing the rotations per second , the two equations may be written\#151 ; n = 12- 4 log I+ 10 , n m 1-56 log I+ 17 *75 , where n is the number of rotations of the disc per second when flicker just vanishes , and I is the illumination of the disc .
The actual " last " ( undiminished ) of the impression on the retina is obtained for any illumination as the fraction of a second by dividing unity by twice n. Thus , for the feeblest illumination experimented with , that of the standard candle at 4 metres , it is about l/ 35th , whilst for the brightest illumination , that of the arc light at 50 cm .
, it is 1/ 147th .
It is also to be noted that the illumination caused by the 25 candle-power lamp at 4 metres is less bright than that of the single candle at half a metre , and also that that of the arc light at 4 metres is as bright as that of the 25 candle-power lamp at half a metre , so that the illuminations given by the moderately bright source of light overlap , as it were , those given by the most feeble and the most intense , and the fact that the values of n obtained were equal when the photometric results showed that they should be so , is a proof that the errors made in estimating the quantitative values of the illuminants were not seriously in error .
Direct photometric comparisons were made between the different illuminants both before and after each set of experiments , and it was thus proved that their illuminating powers had not sensibly changed .
The lowest degree of illumination was one in which , in popular language , it was not easy to see , whilst the brightest was so bright as to make the disc uncomfortable to look at , even when the room in which the experiments were performed was flooded with the diffused light of a bright summer day .
Although the primary object of this research is to throw light upon the process of vision , there are some practical issues to the lines of fig. 3 .
We can easily determine by its means the number of pictures which must be projected on a screen per second in order that there may be no trace of flicker ; the illumination of the brightest part of the brightest view , and that of the darkest part of the darkest view being VOL. LXX .
z 322 Mr. T. 0 .
Porter .
[ May l\#187 ; j known .
and a rough estimate from observations of the brightnesses of the pictures as usually projected by cinematographs on a screen would point to as many as 50 per second being necessary where the arc light is used for projection , though half this number would probably be sufficient to prevent the flicker from being distressing .
This assumes that the photographs are " in register " on the screen .
It has also an important bearing on the usefulness of a " flicker photometer , for if I be the illumination to be measured , and the number of revolutions per second when flicker vanishes , and and M constants , we have , as we have already seen , n \#151 ; k log I + k ' .
Now the sensibility of the photometer depends upon the value of dn/ dl , which , from the above equation is equal to I , that is , it diminishes as I increases , thus the photometer is not adapted for the measurement of brilliant illuminations , at all events directly ; and it must be remembered that , whether sudden or not , if there is a decided change in the value of k in the above expression in the relation between n and I at feeble illuminations , as fig. 2 indicates , the effect of this will be to increase the probable error of a flicker photometer , for k diminishes to practically half its former magnitude , hence the sensi-bility of such a photometer for feeble illuminations will be diminished by about one half .
Finally , if " lbe the " last " of the impression undiminished , measured in this experiment by the time the disc takes to make a half-revolution , ^ == an(l so lis given for the brighter illuminations considered by the equation l - 1/ ( 24-8 log I+ 20 ) , and for the fainter illuminations by l = 1/ ( 3 *12 log I+ 53 *25 ) , the critical illumination being approximately that afforded by the sperm candle already mentioned at a distance of 2 metres .
Experiments were next made to determine how the rate of rotation of a disc part white , part black , alters when the magnitude of the tvhite sector is altered , the illumination of the disc by external sources being maintained constant for each set of experiments .
In giving an account of these experiments and in discussing them it will be necessary to distinguish clearly between ( 1 ) the brightness of the white sector at rest , which is independent of its angular breadth , and depends solely on the distance of the disc from the source of illumination , and on the intrinsic illuminating power of the latter ; and ( 2 ) the apparent brightness of the disc when rotating an less , which depends not only on the external illuminating source ( in fact , j 1902 .
] Contributions to the Study of Fliclcer .
K ( 1 ) ) , but also on the angular breadth of the white'sector itself .
If the words " in sense ( 1 ) " occur hereafter , they refer to the first of the a I^the experiments which have just been described , it will be understood that since the disc remained throughout them half white , half black , its brightness when rotating was never disturbed by any alteration in the width of the white sector , and the results obtained connect " If the " last " of the impression , with I , the brightness of the white sector at rest , or the degree of its illumination by the external source .
In what follows , the illumination afforded by the external illuminant is kept constant for each series of experiments , and the " last " is determined for the various brightnesses of the rotating disc caused by altering the angular magnitude of the white sector ( and necessarily , of course , the angle of the black sector also ) .
The experimental details for the purpose are precisely the same as before described , save that separate black and white discs , dovetailed after Maxwell 's method , were used instead of the single disc , half white half black .
The white disc was gradated about the circumference in lengths subtending at the centre angles of 5 ' , and a concentric siren card mounted on the axis of the same motor , indicated by its note the speed of rotation .
The general character of the results may be , to some extent , foreseen from the results already obtained ; if , as seems likely , the increase of stimulus given by increasing the width of the white sector is essentially of the same nature as that given by increasing *the external illumination , though keeping the white sector constant , we should expect that the connection between u and the angular magnitude of the white sector w ( measured here and elsewhere in degrees ) would be of the form n = / ( log w ) ; and since n = 0 when w * 0 ' or w = 360 ' , n is 7 likely to be a function also of the product w ( 360 - w ) .
Experiment shows that for any constant illumination , using the word in sense ( 1)\gt ; | n = a + , b log w ( 360 - w ) , where a and b are constants , b changes if the illumination ( in sense ( 1 ) ) changes .
The results of the experiments | are most clearly exhibited by the curves in fig. 3 .
Along the axis of X are plotted the notes given by the siren with the number of revolutions per second of the disc to which they correspond .
The ordinates are proportional to the angular magnitude of the white sector , measured in degrees , from 0 ' to 360 ' .
As it was proved that , within the errors !
of experiment , each curve was symmetrical about a line drawn parallel ' to the axis of X through the point corresponding to 180 ' on the axis of Y , only the lower halves of the curves given by experiment are drawn , and these are indicated in continuous lines .
The true origin of these curves lies 55*73 of the paper-scale degrees to the left of the point B corresponding to 20 rotations of the disc per second .
These * five continuous lined curves give the relation ascertained experiment324 Mr. T. C. Porter .
[ May 'flBI ally between n and the varying breadth of the white sector under five different illuminations ( in sense \lt ; 1 ) ) ; these illuminations , being to one another as the numbers 1 , 2 , 4 , 8 , 16 , consequently , by what has gone before , since the logs of these numbers are in A.P. , the distances between points on these five curves , corresponding to the same width , white sector , should lie , as they do , at equal distances measured parallel to the axis of X. The meaning to be attached to the fact that a point \#151 ; say P in the figure on the curve marked 2\#151 ; lies on this continuous lined curve is that , under an illumination called here " 2 " ( actually i that given by a 17*8 candle-power Ediswan lamp at a distance from the rotating disc of 3037 mm. ) , a disc of 110 ' white sector and 250 ' black must be rotated very approximately 32-3 times a second for flicker to vanish .
The dotted curves in the same figure are all derived from the general | expression n = a + hlog w(360 - w ) , by attributing to values which depend on the logs of the different illuminations ( in sense ( 1 ) ) to which !
the disc was exposed in the five experiments by which the experi- j mental curves were obtained .
To make it clear how these curves are drawn , take the last curve on the right for example .
The successive \gt ; products of the numbers expressing the angular magnitudes of corre-1 sponding white and black sectors , e.g. , 10 and 350 , 20 and 340 , 30 and 330 , \amp ; c. , are calculated , and their logs taken j each logarithm is multi- j plied by a factor , b , which is constant for each curve , and which is itself of the form c + dlog I , where c and d are constants , and I is the illumination in sense ?
( 1 ) .
The series of numbers thus obtained is I multiplied by 100 , and each number laid off in the paper-scale !
divisions as units from an origin which lies 132 paper-scale divisions to the left of the point B , corresponding to 20 rotations of the disc per second , and therefore lies 76*5 of the same divisions to the left of the origin of the continuous ( experimental ) curves , and these successive distances are the abscissae of the points which have for ordinates lengths corresponding to the distances marked 10 ' , 20 ' , 30 ' , \amp ; c. , on the axis of Y , being proportional to the angular magnitudes of the white sector in the successive products .
n stands for the number of rotations the disc makes per second when the flicker just vanishes , and it will be found that one division of the paper corresponds to 0*3588 of a rotation per second .
Hence , if is the abscissa of a point P on the experimental curve , reckoned in paper divisions , n = xx 0*3588 or x \#151 ; nl0*3588 .
If X is the abscissa , also in the divisions of the paper , for the same point P on the theoretical ( dotted curves ) , we have x \#151 ; X-76*5 , and therefore X = a:+76 *5 = rc/ 0 * 3588 + 76 * 5 .
But x \#151 ; 100 blog w ( 360 \#151 ; w ) , therefore the equation connecting n and w is nj0*3588 = \#151 ; 76*5 + 100 x x log w ( 360 \#151 ; w ) .
1902.1 Contributions to tfos Study of Flicker .
3-25 The value to be attributed to l must be of the form log I , where t the illumination of the disc for the particular experiment in sense n because this has been already established by the results of the experiments expressed in fig. 3 ; and since the illuminations used to fotein the five curves were 1,21,41 , 81 , and 161 respectively , and I itself .was the illumination of the disc by an incandescent lamp of measured brightness 17-8 ( in terms of the standard candle before mentioned ) at I distance of 3037 mm. from the disc , it will be found that I - 30 9 'units ( the unit being the illumination given by the same candle at a distance from the disc of 4000 mm. ) : c + d log I can therefore he written c + dlog I + mdlog 2 , where in the curves from I XVI on the fig. , m , has the successive values 0 , 1 , 2 , 3 , and 4 .
Log I ipprox .
; c - 0-2560085 ; i - 0-0662837 , and the corresponding values of b become\#151 ; For the curve marked I ... ... ... ... .
0*3547722 XI ... ... ... .
0*3747256 IV 0*3946791 YXXI ... ... ... .
0*4146325 XYI ... ... ... . .
0*4345875 The equation to the curve I is w/ 0*3588 - - 76*5 + 100 x 0*3547722 x log w ( 360 - w ) , with similar equations for the other fTnspeItTon of fig. 4 is sufficient to show how closely the theoretical and experimental curves coincide , but it may be well to take an example or two showing the same thing .
We have seen a expen-1 merit shows that for flicker to vanish on a disc with a white sector of 110 ' under an illumination 2 x 30*9 units , it must be rotated very approximately 32*3 times per second ( vidthe point P , fig. 4 ) .
Now can be calculated from the equation w/ 0*3588 - - 76* + ^ 0*3747256 x log w(360 - w ) by putting w = 110 , when we find n -33*23 .
The experimental value of n 32*3 approx. Simi are y position of the point Q on the curve marked VIII shows that experiment proved that for a disc with a white sector of 130 , under an illumination I = 8 x 30*9 units , a rotation of approximately 39 4 times a second is necessary ; and the result as foundnfr^tQh\#174 ; 'd't^ | curve whose equation is w/ 0*3588 = - 76*5 + 100 x 0*4146325 x log I w(360 - w)by putting w \#151 ; 130 will be found to be 39*13 .
Experiments have proved that each of the experimental curves is \#166 ; symmetrical with respect to the horizontal line passing through the point on the axis of Y marked 180 ' , but only the lower half of the f experimental curves has been drawn .
( This is also true o g. .
) .
\ chief result attained by these experiments is the knowledge of the exact l relation between the variation in the illumination of a disc , having a | white sector , and the number of rotations per second necessary m order ttia f '01 .
Mr. T. C. Porter .
[ May 14( -\lt ; ir-Q\#163 ; -C3 9-63 -U .
ir-93 -uJ 9'9Z -Vr 0-92 -a O-t'S - 9*'39 -o \#163 ; W2 -CO 0*02 - A* 9-9/ ~\lt ; Z-Z/ .* ** Z*9/ - O 0-9/ -Xt 6 \gt ; */ - Ll .
c2*trj - UJ \#163 ; .\#163 ; / - ** 9*2/ -Q 0*2/ - ** tr-H - O / *0/ - CO 0*0/ - ** \#163 ; \gt ; *6 -\lt ; 9-9 Sj^ jwa S\#187 ; \gt ; \gt ; -Q s : .S .So i ?
cu \#163 ; a b !
M Vs ~Tj .
\ Q Ox ^ n * y , ^ ''* N* .
\W * A ^ % % ^ H % , * ' HP V'0'^ 54 \ \\gt ; t ^ 0\ Ifr \#171 ; A f \ '\#171 ; \gt ; r \ t \lt ; o\ ' , \#187 ; ( I -\lt ; -o o % '~7j % Q- -7C % %\gt ; V- v\gt ; \gt ; .\gt ; * \ \ ' , \ ' .
j4 * !
\#166 ; \#171 ; _'\#166 ; '\gt ; tfi| " K @ \ y\ , ' V \gt ; \1\#187 ; v No^j X ' .
ft1 \ \\i A- ^ \ \U ' \lt ; V.iC- ' v '\lt ; * ?
\gt ; * k $'l v \\#171 ; p \\#174 ; Ul | I ' I * l1 -o -**= -oS ai - co -\lt ; \#166 ; '\#163 ; '0'0 ^ ^ Q O '\lt ; 5 !
g jj .
\#174 ; u5 n ) AnguLar magnitude of w. o CJ \#151 ; \lt ; _rj - 4* - Lu - LU ~2 - ** " O - 00 ~Oc ] -co O -** o -\lt ; sN So S3 JjpQ 5'S ^ *o \gt ; ^ '\lt ; 1 CVJ I I* 1902 .
] Contributions to the Study of Flicker .
327 flicker may just vanish .
This , so far as the writer is aware , has never been given before , and it seems remarkable that n should prove a logarithmic function of the illumination ( whether the variation of illu-i niination be due to the external source of light , or whether it be due ito alteration in the magnitude of the white sector , the external I iUuminant remaining constant ) .
It is also of interest to examine | in the light of the results just mentioned the curves expressing ?
the rate at which a disc of variable white sector must be rotated in the different colours of the same spectrum in order that flicker may just vanish .
The figure 3* gives these curves , as found under the conditions there described .
Since the publication of that paper , the experiments there described have been repeated , with the additional precautions mentioned in the present paper , with the result that whilst the general form of the curves has been confirmed , the central part of each , represented in fig. 3 as a vertical straight line , has been proved.^ to be slightly curved , with the concave side facing the axis of Y. These redetermined curves are shown in fig. 4 , for the principal colours of a spectrum ( obtained by means precisely similar to those used in the construction of the curves in the first paper , and which need scarcely be repeated here ) , the dotted curves on the same figure being those derived from x = b log w ( 360 \#151 ; w ) , and y \#151 ; w , where is a constant for each curve , and w is the angular magnitude of the white sector , measured in degrees .
The seven constants for these curves are\#151 ; For the violet ... ... ... ... ... ... .
0*2660459x100 , , blue-violet ... ... ... ... . .
0*2837823x100 " blue ... ... ... ... ... ... . .
0*2993016x100 , , blue-green and crimson . .
0*3148209 x 100 , , full green and vermilion .
0*3303403 x 1Q0 , , yellow-green and scarlet .
0*3547722x 100 " yellow ... ... ... ... ... ... .
0*3747256x100 It will at once be evident that the curves for the different colours coincide , within the errors of experiment , with the dotted curves , and this fact proves that the duration of the impression of the different spectral colours , undiminished , depends solely on the luminosity of the colours , and not on their wave frequency .
This has been suspected before , !
but it has never , so far as the writer knows , been definitely proved .
The proof here given is , that , so far as flicker phenomena go , we obtain exactly the same curves , whether we place the disc in the different colours of the spectrum , or subject it to white light , and * * Roy .
Soc. Proc./ vol. 63 , p. 352 .
f Vide a paper by Professor E. L. Nichols , ' American Journal of Science/ 1884 , No. 28 , pp. 243 to 252 .
328 Contributions to the Study of Flicker .
[ May 14t vary the intensity of its illumination by the white light .
We can , therefore , by the help of what has already been said , deduce from the above constants for each curve in fig. 4 , the illumination of the disc at each point of the spectrum considered , for it follows from the nature of the curves in fig. 3 , that each of the above seven constants is of the form c + dlog I , where c \#151 ; 0*2560085 , and d = 0*066283 , and I is the measure of the illumination to which the disc is exposed in terms of the unit illumination already used .
We have , therefore , 60-i for the violet curve 0*2660459 = 0*2560085 + 0*066283 x log I whence 1= 1*42 approx. Similarly for the other colours in the order of the curves from left to right , we find I *= 2*62 , 4*50 , / *71 , 13 22 , 1902 .
] Refractive Indices of Fluorite , Quartz , and Galcite .
329 30'9 , 61'8 : these values are expressed in the curve of fig. 5 , which is therefore a curve expressing roughly the luminosity of different parts of the same spectrum according to the evidence afforded by flicker .
For the sake of comparison the luminosity of the different colours as measured by Yierordt is indicated by a dotted curve , and it will be seen that the two curves practically coincide , except for the extreme parts of the spectrum ; it should , however , be stated that Vierordt 's curve has been drawn by making his value for the luminosity of the yellow coincide with that given for this colour by the continuous curve in the figure , and reducing his other values in the same proportion : the reduced values are then placed within the limits of the colours to which they are'assigned by Vierordt .
Moreover , Vierordt 's curve is for a solar prismatic spectrum , whilst the other is for incandescent lime , used with a grating.* Lastly it should be remembered that the disc was very feebly illuminated in the violet and extreme crimson , and also , for the small values of the white coloured sector , in the blue , blue-green , and vermilion , so that the alteration of the relation between n and log I ( indicated in fig. 2 by the steeper line for feeble illuminations ) should probably be considered : if this is so , the two curves will be brought into still closer union .
The writer postpones the consideration of this matter , until he has had the opportunity of trying further experiments with such feeble illuminations .
" The Refractive Indices of Fluorite , Quartz , and Calcite .
" By J. William Gifford .
Communicated by Silvanus P. Thompson , F.R.S. Received February 5 , \#151 ; Read February 13 \#151 ; Received in revised form May 7 , 1902 .
1 .
Method of Observation.\#151 ; Measurements of fluorite , quartz , and calcite have been made by Rudberg , Mascart , Cornu , Sarasin , Glazebrook , Van der Willigen , Yogel , Pulfrich , Rubens , Bailey , and many others .
Those now offered were originally undertaken with the view of furthei extending the range and accuracy of lenses constructed of these substances .
A new method of obtaining the refractive indices has been adopted .
Each of the angles of the prisms used was as nearly as possible 60 ' .
When this is the case it is sufficient to measure the deviation of light of a definite wave-length at each angle in turn j the mean of these deviations may be taken as the deviation corresponding * C. Vierordt , ' Pogg .
Ann. , ' yoI .
137 , p. 200 .

