" On the Measurement of the Luminous Intensity of Light .
" By WILLIAM CROOKES , F.R.S. &amp;c. Received June 27 , 1868* .
The measurement of the intensity of a ray of light is a problem the solution of which has been repeatedly attempted , but with less satisfactory results than the endeavours to measure the other radiant forces .
The problem is susceptible of two divisions the absolute and the relative measurement of light .
I. Given a luminous beam , we may require to express its intensity by some absolute term having reference to a standard obtained at some previous time , and capable of being reproduced with accuracy at any time and at any part of the globe .
Possibly two such standards would be necessary , differing greatly in value , so that the space between them might be subdivided into a definite number of equal parts ; or the same result might perhaps be obtained by the well-known device of varying the apparent intensity of the standard light by increasing and diminishing its distance from the instrument .
II .
The standard of comparison , instead of being obtained once for all , like the zeroand boiling-points of a thermometer , may be compared separately at each observation ; and the problem then becomes somewhat simplified into the determination of the relative intensities of two sources of light .
The absolute method is of course the most desirable ; but as the pre** Read December 17 , 1868 : see Abstract , antea , p. 166 .
358 liminary researches and discoveries are yet to be made , before a photometer analogous to a thermometer in fixity of standard and facility of observation could be devised , the realization of an absolute light-measuring method appears somewhat distant .
The path to be pursued towards the attainment of this desirable object appears to be indicated in the observations which from time to time have been made by M. Becquerel , Sir John I-erschel , R. Hunt , and others , on the chemical action of the solar rays , and the production thereby of a galvanic current , capable of measurement on a delicate galvanometer , by appropriate arrangements of chemical baths and metallic plates connected with the ends of the galvanometer wires .
Many so-called photometers have been devised , by which the chemical action of the rays at the most refrangible end of the spectrum have been measured , and the chemical intensity of light tabulated by appropriate methods ; and within the last few years Professors Bunsen and Roscoe have contrived a perfect chemical photometer , based upon the action of the chemical rays of light on a gaseous mixture of chlorine and hydrogen , causing them to combine with formation of hydrochloric acid .
But the measurement of the chemical action of a beam of light is as distinct from photometry proper as is the thermometric registration of the heat-rays constituting the other end of the spectrum .
What we want is a method of measuring the intensity of those rays which are situated at the intermediate parts of the spectrum , and produce in the eye the sensation of light and colour ; and , as previously suggested , there is a reasonable presumption that further researches may place us in possession of a photometric method based upon the chemical action of the luminous rays of light .
The rays which affect an ordinary photographic sensitive surface are so constantly spoken of and thought about as the ultra-violet invisible rays , that it is apt to be forgotten that some of the highly luminous rays of light are capable of exerting chemical action .
Fifteen years ago * the writer was engaged in some investigations on the chemical action of light , and he succeeded in producing all the ordinary phenomena of photography , even to the production of good photographs in the camera , by purely luminous rays of light free from any admixture with the violet and invisible rays .
When the solar spectrum ( of sufficient purity to show the principal fixed lines ) is projected for a few seconds on to a sensitive film of iodide of silver , and the latent image then developed , the action is seen to extend from about the fixed line G to a considerable distance into the ultra-violet invisible rays .
When the same experiment was repeated with a sensitive surface of bromide of silver instead of iodide of silver , the result of the development of the latent image showed that , in this case , the action commenced at about the fixed line 6 , and extended , as in the case of the iodide of silver , far beyond the violet .
A transparent cell , with parallel glass sides one inch across , was filled with a solution of twenty-five parts of sulphate of quinine to one hundred parts of dilute sulphuric acid ; this was placed across the path of the ray of light , and photographs of the spectrum were again taken on iodide of silver and on bromide of silver , the arrangements being , in all cases , identical with those in the first-cited experiments , with the exception of the interposition of the quinine screen .
The action of the sulphate of quinine upon a ray of light is peculiar ; to the eye it scarcely appears to have any action at all , but it is absolutely opaque to the ultra-violet , socalled chemical rays , and thus limits the photographic action on the bromide and iodide of silver to the purely luminous rays .
On developing the latent images , it was now found that the action on iodide of silver was confined to a very narrow line of rays , close to the fixed line G , and in the case of bromide of silver , to the space between b and G. Designating the spaces of action by colours instead of fixed lines , it was thus proved that , behind a screen of sulphate of quinine , iodide of silver was affected only by the luminous rays about the centre of the indigo portion of the spectrum , whilst bromide of silver was affected by the green , blue , and some of the indigo rays .
It is very likely that a continuance of these experiments would lead to the construction of a photometer capable of measuring the luminous rays ; for although bromide of silver behind quinine is not affected by the red or yellow rays , still it is by the green and blue ; and as the proportion of red , yellow , green , and blue rays is always invariable in white light ( or the light would not be white , but coloured ) , a method of measuring one set of the components of white light would give all the information we want-just as in an analysis of a definite chemical compound the chemist is satisfied with an estimation of one or two constituents only , and calculates the others .
Methods based upon the foregoing considerations would supply us with what may be termed an absolute photometer , the indication of which would be always the same for the same amount of illumination , requiring no standard light for comparison ; and pending the development of experiments which the writer is prosecuting in this direction , he has been led to devise a new and , as he believes , a valuable form of relative photometer .
A relative photometer is one in which the observer has only to determine the relative illuminating powers of two sources of light , one of which is kept as uniform as possible , the other being the light whose intensity is to be determined .
It is therefore evident that the great thing to be aimed at is an absolutely uniform source of light .
In the ordinary process of photometry the standard used is a candle , defined by Act of Parliament as a " sperm-candle of six to the pound , burning at the rate of 120 grains per hour .
" This is the standard from which estimates of the value of illuminating gas are deduced ; hence the terms " 12-candle gas , " " 14-candle gas , " &amp;c. In his work on ' Gas Manipulation , ' Mr. Sugg gives a very good account of the difficulties which stand in the way of 360 obtaining uniform results with the Act-of-Parliament candle .
A true sperm-candle is made from a mixture of refined sperm with a small proportion of wax , to give it a certain toughness , the pure sperm itself being extremely brittle .
The wick is of the best cotton , made up into three cords and plaited .
The number of strands in each of the three cords composing the wick of a six-to the-pound candle is seventeen , although Mr. Sugg says there does not appear to be any fixed rule , some candles having more and others less , according to the quality of the sperm .
Sperm-candles are made to burn at the rate of one inch per hour , and the cup should be clean , smooth , and dry .
The wick should be curved slightly at the top , the red tip just showing through the flame , and consuming away without requiring snuffing .
To obtain these results , the tightness of the plaiting and size of the wick require careful attention ; and as the quality of the sperm differs in richness or hardness , so must the plaiting and number of strands .
A variety of modifying circumstances thus tend to affect the illuminating power of a standard sperm-candle .
These difficulties , however , are small compared with those which have resulted from the substitution of paraffin &amp;c. for part of the sperm ; and Mr. Sugg points out that candles can be made with such combinations of stearin , wax , or sperm , and paraffin , as to possess all the characteristics of sperm-candles and yet be superior to them in illuminating-power ; while , on the other hand , candles made from the same materials otherwise combined are inferior .
When , in addition to this , it is found that candles containing paraffin require wicks more tightly plaited and with fewer strands than those suitable for the true sperm-candle , our readers will be enabled to judge of the almost insurmountable difficulties which beset the present system of photometry .
But assuming that the true parliamentary sperm-candle is obtained , made from the proper materials , and burning at the specified rate , its illuminating-power will be found to vary with the temperature of the place where it has been kept , the time which has elapsed since it was made , and the temperature of the room wherein the experiment is tried .
The Rev. W. R. Bowditch , in his work on ' The Analysis , Purification &amp;c. of Coal-gas , ' enters at some length into the question of test-candles , and emphatically condemns them as light-measurers .
One experiment quoted by this author showed that the same gas was reported to be 14'63 or 17'36 candle-gas , according to the way the experiment was conducted .
The present writer has taken some pains to devise a source of light which should be at the same time fairly uniform in its results , would not vary by keeping , and would be capable of accurate imitation at any time and in any part of the world by mere description .
The absence of these conditions seems to be one of the greatest objections to the sperm-candle .
It would be impossible for an observer on the continent , ten or twenty years hence , from a description of the sperm candle now employed , to make a standard which would bring his photometric results into relation 361 with those obtained here .
Without presuming to say that he has satisfactorily solved all difficulties , the writer believes that he has advanced some distance in the right direction , and pointed out the road for further improvement .
Before deciding upon a standard light , experiments were made to ascertain whether the electric current could be made available .
Through a coil of platinum wire , so as to render it brightly incandescent , a powerful galvanic current was passed , and its strength was kept as constant as possible by a thick wire galvanometer and rheostat .
To prevent the cooling action of air-currents , the incandescent coil was surrounded with glass ; and it was hoped that by employing the same kind of battery , and by varying the resistance so as to keep the galvanometer-needle at the same deflection , uniform results could be obtained .
In practice , however , it was found that many things interfered with the uniformity of the results , and the light being much feebler than it was advisable to work with , this plan was deemed not sufficiently promising , and it was abandoned .
The method ultimately decided upon is the following:-Alcohol of sp. gr. 0'805 , and pure benzol boiling at 81 ?
C. , are mixed together in the proportion of 5 volumes of alcohol and 1 of benzol .
This burning fluid can be accurately imitated from description at any future time and in any country ; and if a lamp could be devised equally simple and invariable , the light which it would yield would , it is presumed , be invariable .
This difficulty the writer has attempted to overcome in the following manner .
A glass:lamp is taken of about two ounces capacity , the aperture in the neck being 0'25 inch diameter ; another aperture at the side allows the liquid fuel to be introduced , and , by a well-known laboratory device , the have of the fluid in the lamp can be kept uniform .
The wick-holder consists of a platinum tube 1 81 inch long and 0 125 inch internal diameter .
The bottom of this is closed with a flat plug of platinum , apertures being left in the sides to allow free access of spirit .
A small platinum cup 0'5 inch diameter and 0'1 inch deep is soldered round the outside of the tube 0-5 inch from the top , answering the threefold purpose of keeping the wick-holder at a proper height in the lamp , preventing evaporation of the liquid , and keeping out dust .
The wick consists of fifty-two pieces of hard-drawn platinum wire , each 0-01 inch in diameter and 2 inches long , perfectly straight , and tightly pushed down into the platinum holder , until only 01 inch projects above the tube .
The height of the burning fluid in the lamp must be sufficient to cover the bottom of the wickholder : it answers best to keep it always at the uniform distance of 1175 inch from the top of the platinum wick ; a slight variation of level , however , has not been found to influence the light to an extent appreciable by our present means of photometry .
The lamp with reservoir of spirit thus arranged , with the platinum wires parallel , and their projecting ends level , a light is applied , and the flame instantly appears , forming a perfectly shaped cone 1'25 inch in height , the point of maximum brilliancy being 0'56 inch from the top of the wick .
The extremity of the flame is perfectly sharp without any tendency to smoke ; without flicker or movement of any kind , it burns when protected from currents of air at a uniform rate of 136 grains of liquid per hour .
The temperature should be about 60 ?
F. , although moderate variations on either side exert no perceptible influence .
Bearing in mind Dr. Frankland 's observations on the direct increase in the light of a candle with the atmospheric pressure , accurate observations ought to be taken only at one height of the barometer .
To avoid the inconvenience and delay which this would occasion , a table of corrections should be constructed for each 01 inch variation of barometric pressure .
There is no doubt that this flame is very much more uniform than that of the sperm-candle sold for photometric purposes .
Tested against a candle , considerable variations in relative illuminating-power have been observed ; but on placing two of these lamps in opposition , no such variations have been detected .
The same candles have been used , and the experiments have been repeated at wide intervals , using all customary precautions to ensure uniformity .
The results are thus shown to be due to variations in the candle , and not in the lamp .
It is expected that whoever may be inclined to adopt the kind of lamp here suggested will find not only that its uniformity may be relied upon , but that , by following accurately the description and dimensions here laid down , each observer will possess a lamp of equivalent and convertible photometric value ; so that results may not only be strictly comparable between themselves , but , within slight limits of accuracy , comparable with those obtained by other experimentalists .
The dimensions of wick &amp;c. here laid down are not intended to fix the standard .
Persons engaged in photometry as an important branch of their regular occupation will be better able to fix these data than the writer , by whom photometry is only occasionally pursued as a means of scientific research .
Already many improvements suggest themselves , and several causes of variation in the light have been noticed .
Future experiments may point out how these sources of error are to be overcome ; but at present there is no necessity to refine our source of standard light to a greater degree of accuracy than the photometric instrument admits of .
The instrument for measuring the relative intensities of the standard and other lights next demands attention .
The contrivances in ordinary use are well known .
Most of them depend on the law in optics , that the amount of light which falls upon a given surface varies inversely with the square of the distance between the source of light and the object illuminated .
The simplest observation which can be taken is made by placing two sources of light ( say , a candle and gas-lamp ) opposite a white screen a few feet off , and placing a stick in front of them , so that two shadows of the stick may fall on the screen .
The strongest light will cast the strongest shadow ; and by moving this light away from the stick , keeping the shadows side by side , a position will at last be found at which the two shadows appear of equal strength .
By measuring the distance of each light from the screen and squaring it , the product will give the relative intensities of the two sources of light .
In practice this plan is not sufficiently accurate to be used except for the roughest approximations ; and from time to time several ingenious contrivances , all founded upon the same law , have been introduced by scientific men by which a much greater accuracy is obtained ; thus , in Ritchie 's photometer , the lights are reflected on to a piece of oiled paper in a box , and their distances are varied until the two halves of the paper are equally illuminated .
In Bunsen 's photometer , which is the one now generally used , the lights shine on opposite sides of a disk of white paper , part of which has been smeared with melted spermaceti to make it more transparent .
When illuminated by a front light , the greased portion of the paper will look dark ; but if the observer goes to the other side of the paper , the greased part looks the lighter .
If , therefore , lights of unequal intensity are placed on opposite sides of a piece of paper so prepared , a difference will be observed ; but by moving one backwards or forwards , so as to equalise the intensity , the whole surface of the paper will appear uniformly illuminated on both sides .
This photometer has been modified by many observers .
By some the disk of paper is moved , the lights remailing stationary ; by others the whole is enclosed in a box , and various contrivances are adopted to increase the sensitiveness of the eye , and to facilitate calculation : but in all these the sensitiveness is not greatly augmented , as the eye cannot judge of very minute differences of illumination approximating to equality .
In 1833 Arago described a photometer in which the phenomena of polarized light were employed .
This instrument is fully described , with drawings , in the tenth volume of the ' ( Euvres completes de Frangois Arago ; ' but the description , although voluminous , is far from clear .
The principle of its construction is founded on the law of the square of the cosines , according to which polarized rays pass from the ordinary to the extraordinary image .
The knowledge of this law , he says , will not only prove theoretically important , but will further lead to the solution of a great number of very important astronomical questions .
Suppose , for example , that it is wished to compare the luminous intensity of that portion of the moon directly illuminated , by the solar rays , with that of the part which receives only light reflected from the earth , called the party cendree .
Were the law in question known , the way to proceed would be as follows : After having polarized the moon 's light , pass it through a doubly refracting crystal , so disposed that the rays , not being able to bifurcate , may entirely undergo ordinary refraction .
A lens placed behind this crystal will therefore show but one image of our satellite ; but as the crystal , in rotating on its axis , passes from its original position , the second image will appear , and its intensity will go on augmenting .
The movement of the crystal must be arrested at the moment when , in this growing extraordinary image , the segment corresponding to the part of the moon illuminated by the sun exhibits the intensity of the ashy part shown by the ordinary image .
From these data it is easy to perceive , he says , that the problem is capable of solution .
In another part of the same volume , after speaking of the polariscope which goes by his name , Arago writes:- " I have now arrived at the general principle upon which my photometric method is entirely founded .
The quantity ( I do not say the proportion ) the quantity of completely polarized light which forms part of a beam partially polarized by reflection , and the quantity of light polarized rectangularly which is contained in the beam transmitted under the same angle , are exactly equal to each other .
The reflected beam , and the beam transmitted under the same angle by a sheet of parallel glass , have in general very dissimilar intensities ; if , however , we examine with a doubly refracting crystal first the reflected and then the transmitted beam , the greatest difference of intensity between the ordinary and the extraordinary images will be the same in the two cases , because this difference is precisely equal to the quantity of polarized light which is mixed with the common light .
" In Arago 's 'Astronomy , ' the author again describes his photometer in the following words:- " I have constructed an apparatus by means of which , upon operating with the polarized image of a star , we can succeed in attenuating its intensity by degrees exactly calculable after a law which I have demonstrated .
" It is difficult to obtain an exact idea of this instrument from the description given ; but from the drawings it would appear to be exceedingly complicated and to be different in principle and construction from the one now about to Fig. be described .
The present photometer has this in common with that of Arago , as well as with those deD c scribed in 1853 by Bernard * , and in 1854 by Babinet + , that the phenomena of polarized light are used for effect ing the desired end ; but it is believed that the present arrangement is quite new , and it certainly appears to answer the purpose in a way which leaves little to be e/ desired .
The instrument will be better understood if )(ID the principles on whiGh it is based are first described .
Fig. 1 shows a plan of the arrangement of parts , not .
-drawn to scale , and only to be regarded as an outline sketch to assist in the comprehension of general principles .
Let D represent a source of light .
This may be a white disk of porcelain or paper illuminated by any artificial or natural light .
C represents a similar ( white disk , likewise illuminated .
It is required to com* Comptes Rendus , April 25 , 1853 .
t Proceedings of the British Association , Liverpool Meeting , 1854 .
365 pare the photometric intensities of D and C. ( It is necessary that neither D nor C should contain any polarized light , but that the light coming from them , represented on each disk by the two lines at right angles to each other , forming a cross , should be entirely unpolarized .
) Let H represent a double refracting achromatic prism of Iceland spar ; this will resolve the disk D into two disks d and d ' , polarized in opposite directions ; the plane of d being , we will assume , vertical , and that of dt horizontal .
The prism HI will likewise give two images of the disk C ; the image c being polarized horizontally , and c ' vertically .
The size of the disks D , C , and the separating power of the prism H , are to be so arranged that the vertically polarized image d , and the horizontally polarized image c , exactly overlap each other , forming , as shown in the figure , one compound disk c d , built up of half the light from D and half that from C. The measure of the amount of free polarization present in the disk cd will give the relative photometric intensities of D and C. The letter I represents a diaphragm with a circular hole in the centre , just large enough to allow the compound disk cd to be seen , but cutting off from view the side disks c ' , d ' .
In front of the aperture in I is placed a piece of selenite , of appropriate thickness for it to give a strongly contrasting red and green image under the influence of polarized light .
K is a doubly refracting prism , similar in all respects to H , placed at such a distance from the aperture in I that the two disks into which I appears to be split up are separated from each other , as at g , r. If the disk cd contains no polarized light , the images g , r will be white , consisting of oppositely polarized rays of white light ; but if there is a trace of polarized light in c d , the two disks g , r will be coloured complementarily , the contrast between the green and the red being stronger in proportion to the quantity of polarized light in c d. The action of this arrangement will be readily evident .
Let it be supposed , in the first place , that the two sources of light , D and C , are exactly equal .
They will each be divided by I into two disks d ' d and c ct , and the two polarized rays of which cd is compounded will also be absolutely equal in intensity , and will neutralize each other , and form common light , no trace of free polarization being present .
In this case the two disks of light , g , r , will be colourless .
Let it now be supposed that one source of light ( D , for instance ) is stronger than the other ( C ) .
It follows that the two images d ' , d will be more luminous than the two images c , cr , and that the vertically polarized ray d will be stronger than the horizontally polarized ray c. The compound disk cd will therefore shine with partially polarized light , the amount of free polarization being in exact ratio with the photometric intensity of D over C. In this case the image of the selenite plate in front of the aperture I will be divided by K into a red and a green disk .
Fig. 2 shows the instrument fitted up .
A is the eyepiece ( shown in enlarged section at fig. 3 ) ; GB is a brass tube , blacked inside , having a 366 piece ( shown separate at D C ) slipping into the end B. The sloping sides , D B , B C , are covered with a white reflecting surface ( white paper or finely Fig. 2 .
_ I ?
ground porcelain ) , so that when DC is pushed into the end B , one white surface DB may be illuminated ( as in fig. 2 ) by the candle , and the other surface BC by the lamp , If the eyepiece A is removed , the observer , looking down the tube G B , will see at the end a luminous white disk divided vertically into two parts , one half being illuminated by the candle E , and the other half by the lamp F. By moving the candle E , for instance , along the scale , the illumination of the half DB carl be varied at will , the illumination of the other half remaining stationary .
The eyepiece A ( shown enlarged at fig. 3 ) will be understood by reference to fig. 1 , the same letters representing similar ig .
3 .
parts .
At L is a lens to collect the rays from DBCL ( fig. 2 ) , and throw the image into the proper part | of the tube .
At M is another lens so adjusted as to give a sharp image of the two disks into which I is divided by the prism K. The part N is an adaptation of Arago 's polarimeter ; it consists of a series of thin plates of glass capable of moving round the axis of the tube , and furnished with aI pointer and graduated arc ( shown at A G , fig. 2 ) .
By means of this pile it is possible to partially polarize the rays coming from the illuminated disks in one or the other direction , and thus bring to the neutral state the partially polarized beam cd ( fig. 1 ) , i so as to get the images g , r free from colour .
It is so adjusted that when at the zero-point it produces an : K equal effect on both disks .
The action of the instrument is as follows .
The M --:--standard lamp being placed on one of the supporting pillars which slide along the graduated stem ( fig. 2 ) , it is adjusted to the proper height , and moved 367 along the bar to a convenient distance , depending on the intensity of the light to be measured : the whole length being a little over 4 feet , each light can be placed at a distance of 24 inches from the disk .
The flame is then sheltered from currents of air by black screens placed round , and the light to be compared is fixed in a similar way on the other side of the instrument .
The whole should be placed in a dark room , or surrounded with nonreflecting screens ; and the eye must also be protected from the direct rays of the two lights .
On looking through the eyepiece two bright disks will be seen , probably of different colours .
Supposing F represents the standard flame , and E the light to be compared with it , the latter must now be slid along the scale until the two disks of light , seen through the eyepiece , are about equal in tint .
Equality of illumination is easily obtained ; for as the eye is observing two adjacent disks of light , which pass rapidly from redgreen to green-red through a neutral point of no colour , there is no difficulty in hitting this point with great precision .
It has been found most convenient not to attempt to get absolute equality in this manner , but to move the flame to the nearest inch on one side or the other of equality .
The final adjustment is now effected at the eye-end by turning the polarimeter one way or the other up to 45 ?
, until the images are seen without any trace of colour .
This will be found more accurate than the plan of relying entirely on the alteration of the distance of the flame along the scale ; and by a series of experimental adjustments the value of every angle through which the bundle of plates is rotated can be ascertained once for all , when the future calculations will present no difficulty .
Squaring the number of inches between the flames and the centre will give their approximate ratios ; and the number of degrees the eyepiece rotates will give the number to be added or subtracted in order to obtain the necessary accuracy .
The delicacy of the instrument is very great .
With two lamps , each about 24 inches from the centre , it is easy to distinguish a movement of one of them to the extent of 0'1 inch to or fro ; and by using the polarimeter an accuracy considerably exceeding this can be attained .
The employment of a photometer of this kind enables us to compare lights of different colours with one another , and leads to the solution of a problem which , from the nature of their construction , would be beyond the powers of the instruments in general use .
So long as the observer , by the eye alone , has to compare the relative intensities of tint-surfaces , respectively illuminated by the lights under trial , it is evident that , unless they are of the same tint , it is impossible to obtain that equality of illumination in the instrument which is requisite for a comparison .
By the unaided eye one cannot tell which is the brighter half of a paper disk , illuminated on one side with a reddish , and on the other with a yellowish light ; but by using the above-described photometer the problem becomes practicable .
For instance , on reference to fig. 1 , suppose the disk D were illuminated with light of a reddish colour , and the disk C with greenish 368 light , the polarized disks d ' , d would be reddish and the disks c , c1 greenish , the central disk cd being of the tint formed by the union of the two shades .
The analyzing prism K , and the selenite disk I , will detect free polarization in the disk c d , if it be coloured , as readily as if it were white ; the only difference being that the two disks of light , g , r , cannot be brought to a uniform white colour when the lights from D and C are equal in intensity , but will assume a tint similar to that of c d. When the contrasts of colour between D and C are very strong-when , for instance , one is bright green and the other scarlet-there is some difficulty in estimating the exact point of neutrality ; but this only diminishes the accuracy of the comparison , and does not render it impossible , as it would be according to other systems .
No attempt has been made in these experiments to ascertain the exact value of the standard spirit-flame in terms of the Parliamentary spermcandle .
Difficulty was experienced in getting two lots of candles yielding light of equal intensities ; and when their flames were compared between themselves and with the spirit-flame , variations of as much as 10 per cent. were sometimes observed in the light they gave .
Two standard spiritflames , on the other hand , seldom showed a variation of 1 per cent. , and had they been more carefully made they would not have varied 01 per cent. This plan of photometry is capable of far more accuracy than the present instrument will give .
It can scarcely be expected that the first instrument of the kind , made by an amateur workman , should possess equal sensitiveness with one in which all the parts have been skilfully made with special adaptation to the end in view .

