245 A Bolometric Method of Determining the Efficiencies of Radiating Bodies .
By William A. Bone , F.B.S. , H. L. Callendar , F.B.S. , and H. James Yates .
( Beceived February 15 , 1915 .
) Introduction .
With the increasing use , within recent years , of gas fires , electric radiators , incandescent surface combustion diaphragms , and the like , there has arisen a demand for a reliable general method for testing their radiant efficiencies .
The problem is important also in its purely scientific aspects , inasmuch as its solution would enable the relations between the modes of combustion of various combustible gases and their radiant values when burnt at an incandescent surface to be investigated .
The present paper describes a bolometric method which we believe to be applicable to the measurement of the radiant energy emitted from hot terrestrial surfaces generally , but as the source of radiation in our experiments has been a gas fire , it will be convenient if we describe it with reference to that particular mode of heating .
A typical modern gas fire consists of a series of atmospheric burner nozzles , varying usually between seven and fifteen in number , according to the size of the apartment to be heated , arranged equidistantly along a common horizontal supply pipe provided with suitable means for properly adjusting the relative gas and primary air supplies so as to ensure a vertical series of non-luminous and , as nearly as possible , silent Bunsen flames , of uniform height and character .
Above each flame is fixed a hollow fireclay columnal " radiant " perforated in a manner expressly contrived to promote uniform heating of the column throughout , and with each flame rising into the cavity of its particular radiant .
Care is taken to prevent any impingement of the inner cone of the flame upon its radiant .
The back of the fire is formed by a fireclay slab fixed vertically behind the radiants , which are held in position by means of one slight horizontal iron rod .
At a suitable distance above the top of the radiants is fixed a metallic hood or " canopy " by means of which the products of combustion are collected and discharged through the flue vent into the chimney without in any way contaminating the atmosphere of the apartment .
The object of a properly designed fire should be to secure the maxima of radiant and ventilating effects with a minimum of " flue heat/ ' and it is in connection with these aspects of the problem that the need of convenient and 246 Messrs. Bone , Callendar , and Yates .
A Bolometric accurate means of measuring the radiant efficiency of the fire has been mostly felt.* # ' .
A joint committee appointed in 1907 by the University of Leeds and the Institution of Gas Engineers to investigate gas fires adopted a method ( now known as the Leeds method ) for determining their radiant efficiencies .
It consists essentially in firstly establishing , by thermopile readings , a relation between the intensity of the radiation at a central equatorial area on a hemisphere in front of the fire , and the total radiation received over the whole hemisphere ( a relation which will be hereafter referred to as the " distribution factor " of the fire ) , and then determining , by means of a radiometer , of the water-calorimeter type , designed by Prof. B. H. Smith , the actual number of calories radiated by the fire per hour on to the said equatorial area .
The number of such calories multiplied by the " distribution factor " of the fire gives the total energy radiated by the fire per hour , and the relation of this to the total net heat developed by the combustion of the gas in the fire per hour gives the radiant efficiency of the fire.f This method , whilst perfectly sound in principle , is admittedly open to criticism on the following grounds , namely :\#151 ; ( 1 ) The B. H. Smith radiometer is liable to various small indeterminable errors inherent in all such water-calorimeter radiometers , due principally to ( a ) imperfect absorption of the radiation by its blackened surface , ( b ) effects / of " lag , " which , however , if the experimental conditions remain constant , may be disregarded , and ( c ) the difficulty of carrying out a really satisfactory " blank " experiment to determine the allowance to be made in the actual test for heat gained by the instrument from the surrounding warmer atmosphere.^ ( 2 ) It is practically impossible , owing to the large area of the absorbing surface of such a radiometer , to standardise its readings from a known absolute radiation standard .
( 3 ) The time taken to complete both the radiometer ( actual and blank ) tests and the 81 thermopile readings required to establish the " distribution * See also H. James Yates on " Beeent Progress in Gas-Fire Science , " 1 Brit. Assoc. Beports , ' Birmingham , 1913 , pp. 435-9 .
t A full description of the Leeds method is given in the Committee 's First and Second Beports , ' Trans. Inst. Gas Engineers , ' 1909 , pp. 59-81 .
X There is certainly a small systematic error in the " blank " test , due to the impossibility of screening the radiometer perfectly from the fire during the blank , the effect of which is to make the ascertained " blank correction " too high .
By means of the bolometer we have estimated that , in the case of a 10-inch gas fire , and using a single reflecting screen in front of the radiometer during the Leeds blank test , the error may amount to T)*44 kilogramme-calories per hour on a total centre reading ( in the actual test ) of 55-60 kilogramme-calories .
Method of Determining the Efficiencies Radiating Bodies .
247 factor " is considerable , and there is always a risk of some alteration during the test in the experimental conditions which would affect its result .
( 4 ) There is also perhaps a little uncertainty about the absolute accuracy of the " distribution factor " as determined by the Rubens thermopile used in the Leeds method .
In any case , were the method entirely free from any or all such practical objections , it would still be important to compare its results with those obtained by some independent standard method .
The New Bolometric Method .
The problem of measuring the radiant efficiency of a modern gas fire , such as has been described in the introductory part of our paper , is complicated by the fact that the fire front ( .
, the seat of the radiation ) cannot be regarded as a surface of simple geometric form .
The seat of the radiation has , in fact , a measurable depth , and , owing to the line of the fire front being often slightly curved , its exterior surface is not always flat .
Again , the hemispherical distribution of the radiation may be disturbed either by absorption or by reflection from projecting surfaces on the casing of the fire .
It is necessary to determine , not only the whole radiation emitted by a particular fire , but also its distribution factor , because a proper distribution of the radiant energy is almost as important as its total amount .
The Bolometer and its Advantages .
We propose to substitute for the radiometer-cum-thermopile device in the Leeds method a simple bolometer in which the radiation falling from the fire upon a blackened coil of platinum wire can be deduced from the observed increase in its electrical resistance , the area of the receiving coil being sufficiently small to allow of the instrument being standardised directly from a source of radiation of known intensity .
The principal object in employing a bolometer in place of a thermopile is to secure a greater range of accuracy and sensitiveness , and to facilitate the obtaining of automatic records when required .
The sensitiveness of a bolometer is readily varied over a wide range by varying the electric current employed for measuring the resistance .
The sensitiveness may also be determined very easily , under any conditions , by observing the deflection produced by inserting a known resistance in the circuit .
In using the Rubens thermopile ( as in the Leeds method ) , it is found necessary to attach a reflecting cone to the instrument in order to obtain a deflection of 25 scale divisions with a Paul unipivot galvanometer at a distance of 248 Messrs. Bone , Callendar , and Yates .
A Bolometric 3 feet from a 10-inch gas fire .
The use of a reflecting cone narrows the angular aperture , and introduces some uncertainty with regard to the extent of the source from ' which the radiation is actually received .
The 'bolometer , when supplied with the current for which it is designed , and employed in conjunction with the same unipivot galvanometer but without any reflecting cone , is found to have a sensitiveness about thirty times as great as the Eubens thermopile with the cone , and could be used at much greater distances from the fire .
By adjusting the sensitiveness to a suitable figure in scale divisions per ohm , it is always possible to obtain deflections in the neighbourhood of 90 or 100 scale divisions , which greatly facilitates the accurate determination of the radiation .
Construction of the Bolometer .
The familiar type of bolometer invented by Prof. Langley has a sensitive receiving surface in the form of a grid cut from thin metal foil , and is blackened with smoke black or platinum black .
This method of construction was adopted with the object of securing the greatest quickness of action , but it involves extreme fragility and is wanting in constancy .
The coating of black invariably deteriorates , and cannot be renewed without altering the constant of the instrument .
Constancy is much more important for the present purpose than quickness of action , and fragility would be a serious defect .
A different method of construction was accordingly adopted in our experiments ; the delicate grid of the Langley bolometer has been replaced by a coil of platinum wire wound on a thin piece of mica , and coated to an even surface with hard black enamel , which is extremely permanent , and can be cleaned without risk of injury .
Two exactly similar coils , each 4 cm .
square and of 20 ohms resistance , are mounted back to back on either side of a circular gunmetal box , provided with water circulation , and with suitable covers for the coils , so that either coil can be exposed to radiation or screened .
When both coils are screened they are kept at the same temperature as the box , their resistances remain equal , and there is no deflection of the galvanometer , however much the temperature of the box may change .
But if one of fcthe coils is exposed to radiation , its temperature and resistance are increased by an amount depending on the intensity of the radiation , and the galvanometer shows a deflection proportional to the increase of resistance , which serves as a measure of the intensity of the incident radiation .
Method of Determining the Efficiencies of Radiating Bodies .
249 Calibration of the Bolometer .
An instrument of this type can be directly calibrated to give the intensity of the radiation in absolute measure , by observing the magnitude of the electric current required to produce the same rise of temperature , or increase of resistance , as the radiation to be measured .
This method has often been adopted , but is not quite satisfactory , on account of the uncertainty of the absorption factor of the surface for radiation , which is one of the commonest sources of error in all measurements of radiation .
The constant of the bolometer , giving the intensity of the radiation in terms of increase of resistance , was accordingly determined by comparison with a radio-balance , an instrument specially designed to give total absorption of the radiation.* In the radio-balance , the radiation to be measured is admitted through a small circular aperture and received in a blackened copper cup , in which the absorption is practically complete .
The heat received from the radiation is balanced by absorption of heat due to the Peltier effect in a thermojunction through which a measured electric current is passed .
This method is very accurate , but is not well suited for ordinary use outside a physical laboratory , because it requires a 'sensitive galvanometer for indicating the balance , and a delicate ammeter or potentiometer for measuring the electric current .
Nevertheless , a bolometer such as the one herein described can , without risk of fracture of its working parts , be sent to a physical laboratory for comparison with a radio-balance .
The following Table contains all the results for the constant of this particular bolometer obtained by comparison with two different radiobalances , denoted by I ) and E respectively .
Two different sources of Observations of Bolometer Constant K. Date .
Radio -balance .
Temperature of bolometer .
Air .
Source of radiation .
Constant of bolometer .
1914 .
July 25 E o 21 -5 o 22 -0 Gias fire 27 -71 " 26 E 21 *0 21 5 Focus lamp Gras fire 27 '85 Nov. 28 D 15 *7 18 *0 28 -09 " 29 D 17 2 18*7 Focus lamp Gas fire 28 -02 Dec. 1 D 17 *7 18 -0 27 '82 " 3 E 18 -9 18 *2 Focus lamp 27 96 " 3 D 18*7 18 *3 27 '86 " 5 E 20 *7 19 *2 Gas fire 27 '82 " 6 E 18*3 21 *2 .
27 '78 Means 18 *9 19 *4 K = 27 '88 * ' Proc. Phys. Soc. Lond. , ' vol. 23 , pp. 1-34 ( 1910 ) .
250 Messrs. Bone , Callendar , and Yates .
A Bolometric radiation were employed\#151 ; ( 1 ) a 10-inch gas fire at a distance of 1 metre , ( 2 ) a focus lamp of 100 candle power , having a small radiant 1 inch square , at a distance of 30 cm .
It was unfortunately impossible to vary the temperature of the laboratory materially , but the temperature of the bolometer was altered 5 ' or 10 ' on each occasion , by changing the water circulation , in order to determine the correction for the difference of temperature between the bolometer and the air .
The results given for the constant are corrected for the observed difference .
The value of the constant is given as the intensity of the radiation in kilocalories per square foot per hour required to produce an increase of resistance of 1 ohm in the exposed coil of the bolometer .
The chief source of error in the comparisons was the uncertainty of temperature of the surrounding air and the walls of the room , which cannot be well avoided in measurements of this kind with a source of large area like a gas fire , since the receiving instrument is necessarily exposed to the air and to radiation from surrounding objects , which may be at different temperatures .
The correction for difference of temperature between the air and the bolometer was found to be 0*028 ohm per 1 ' C. , and amounted in some cases to as much as 4 per cent. , when the difference of temperature was considerable , but the divergence of the corrected results from the mean in no case reaches 1 per cent. , and is generally less than 0*5 per cent. The Distribution Factor .
The " distribution factor " of a gas fire , or other approximately flat source of radiation , may be defined as the factor by which the normal intensity at a given distance D must be multiplied in order to obtain the total radiation emitted over a hemisphere .
If the source is a plain circular disc of uniform intensity and of radius B , the value of the factor given by the theory of radiation for a self-luminous surface , would be simply 7t(D2 + B2 ) .
Thus , for example , in the case of a plane source 1 foot in diameter , tested at a distance of 9/ nr feet , or 34#38 inches , the value of the distribution factor should be 26*58 , which is about the value generally obtained for a 10-inch gas fire , tested at this distance , by taking a large number of readings equally distributed over the surface of a hemisphere .
In the case of a gas fire the surface is , however , seldom flat , and there are considerable extensions , such as the canopy , which contribute an appreciable fraction of the radiation .
The chief uncertainty arises in estimating the distance D. An error of 1 inch in^D in the above example would make an error of nearly 6 per cent , in the mathematical formula , and it may be doubted whether this error Method of Determining the Efficiencies of Radiating Bodies .
251 can be eliminated in practice by taking observations over a hemisphere of fixed radius .
The obvious method of reducing this uncertainty , especially with a large gas fire , would be to increase the distance .
But this introduces some practical difficulties if a large number of readings are to be taken over a hemisphere of large radius .
It is doubtful whether an accurate result can be obtained by mere multiplication of readings .
Since the deviation of the distribution factor from the theoretical value for a flat surface is evidently small , it would probably suffice to take a few readings at suitably selected angles .
It appears from theory , and also from comparison of observations , that the sum of four readings taken at 60 ' along the equator and meridian , together with the central reading , multiplied by ' ?
rD2/ 3 , ( 8*6 , when D = 9/ 7t feet ) , gives as good an approximation to the total radiation as can be obtained by taking 81 .
readings .
The distribution factor is represented by the sum of the five readings divided by the central reading and multiplied by 7tD2/ 3 .
Effect of Variation of Temperature of Receiver .
The importance of keeping the temperature of the receiver , whether bolometer or thermopile , constant , is best appreciated by taking a case in which this precaution is omitted .
If a Rubens thermopile with a Paul indicator ( such as is used in the Leeds method ) is exposed to a steady source , giving an intensity of 46 kilocalories per square foot per hour , the deflection rises to about 25 mm. in 15-20 seconds , and creeps up to 27*5 , a further 10 per cent. , in three or four minutes .
The deflection then diminishes to about 25 mm. if the exposure is prolonged , on account of rise of temperature of the case , which may amount to 5 ' or 10 ' C. If , now , the hot instrument is screened from the source , or turned towards the wall of the room , which is at a lower temperature , it falls quickly to zero and gradually takes up a negative deflection of 3 or 4 mm. In other words , the rise of temperature of the instrument may produce an effective change of zero equivalent to 10 or 15 per cent , of the deflection .
Variations of this magnitude make it impossible to employ the pile with a constant reduction factor for determining the absolute value of the radiation .
In using the pile merely for the distribution factor , the errors are not so obvious , but , being systematic , they produce a marked effect on the results , as is shown by the following typical comparison of readings taken with thermopile and bolometer , along the equator in front of a gas fire .
vol. xci.\#151 ; A. x Messrs. Bone , Callendar , and Yates .
A Bolometric Position 60 ' West .
0 ' .
60 ' East .
Thermopile 16 *0 23 *3 13 *0 Bolometer 58 *2 91 *2 58 *0 Thermopile 15 *2 22 *9 13 0 The thermopile readings make the distribution appear about 20 per cent , greater on the west than on the east .
There is no reason why this should be the case , and the bolometer readings show practically no difference between east and west .
The explanation is that in the case of the thermopile the west readings were taken first , while the deflection was still increasing , whereas the east readings were taken after the instrument had become hot , with the consequent depression of zero .
The opposite errors on the two sides tend to neutralise each other to some extent , but the general effect is to reduce all the smaller readings unduly as compared with the central reading , so that the distribution factor obtained with the thermopile may be as much as about 3-5 per cent , too small .
This point may be illustrated by reference to the following Table of the " Distribution factor " of the same 10-inch gas fire as determined on several different occasions by means of ( 1 ) the bolometer , and ( 2 ) a Rubens thermopile .
The Distribution Factor of a 10-inch Gas Fire as determined by\#151 ; Bolometer .
Thermopile .
28-23 27-42 2856 27-55 28-65 27-12 28-30 27-55 28-70 27-57 28-16 27-68 27-97 27-52 28-52 27-81 27-13 28-52 Mean = 27*44 28-47 Ratio _ ^'36 _ 1-034 .
Mean = 28-36 27-44 Method of Testing a Gas Fire .
It is usually the most convenient method , in testing a gas fire , to mount the bolometer or thermopile on a revolving sector of fixed radius , the centre of Method of Determining the Efficiencies of Radiating Bodies .
253 which is adjusted to coincide approximately with the centre of the fire .
If the radius of the hemisphere over which the bolometer moves is 9/ 7t feet , or 34*4 inches , as employed in the Leeds method , an error of 1 inch in the adjustment of the sector along the normal will produce an error of 6 per cent , in the central reading .
It is difficult to specify accurately how this adjustment should be made , especially if a radiometer of large receiving surface , such as 1 sq .
ft. , is employed for the central reading .
The Leeds method of testing throws too much weight on the central reading , which is the most uncertain so far as the adjustment is concerned .
The reasons for taking four readings at an angle of 60 ' with the normal are ( 1 ) that these readings take very fair account of the variation of distribution in different directions , ( 2 ) that the sum of the four readings will be very nearly independent of small errors of adjustment of the sector , and will reduce to about one-third the probable error of adjustment on the central reading .
Automatic Records .
It is possible with the bolometer to obtain automatic records in ink on a large scale , which are useful in recording the progress of a test , or showing the effect of variation of conditions .
The scale is readily varied from 10 to 40 cm .
per ohm .
The reading obtained with a 10-inch gas fire at a distance of 3 feet is generally between 1*5 and 2 ohms with the bolometer already described .
This reading would correspond to a vertical height of 15 to 80 cm .
on the record according to the scale employed .
The smaller scale is quite sufficient for an ordinary test , but the larger scale is useful in investigating small corrections , due to variation of temperature of the room or similar causes .
Leadings can be taken more quickly with an indicator such as the unipivot galvanometer already mentioned , but the recorder has the advantage that its scale is more accurate and uniform , and that it is easier to see when r the conditions of observation are satisfactorily steady .
Radiant Efficiency of a Gas Fire .
In conclusion , we may tabulate the results of a series of determinations of the radiant efficiency of a 10-inch gas fire working on London coal gas of average net calorific value 32*88 K.C.Us .
per cubic foot at N.T.P. , with an average consumption of 24*25 cu .
ft. per hour at N.T.P. , supplied to the fire at a constant pressure of 1*9 inches ( water gauge ) above the barometric pressure .
Determination of Efficiencies of Radiating Bodies .
Date .
Barometer .
Dry gas consumption per hour at N.T.P. Net calorific value dry gas K.C.U. 's per cub. ft. at N.T.P. \#171 ; Radiant efficiency .
1914 .
mm. cub. ft. per cent. Dec. 9 750 -3 24 *00 132 *0 44 *8 " 10 752 -5 24 *49 131 0 45 -0 " 11 746 -1 24 *25 130 *6 43 *2 " 14 732-7 23 *78 130 *2 46 -0 " 16 748 -4 24 *19 131 *8 46 *6 " 17 761 -1 24 *77 132 *3 44 -6 764 -7 24 -49 132 *3 44 -6 763 -2 24 *51 131 *7 45 5 763 -2 24 *26 131 *7 45 *5 " 18 751 -4 23 *82 130 *6 44 *0 Mean 45 -0 The testing room was a laboratory 38 ft. x 32J ft. x 15 ft. high , the temperature of which varied only between 14 ' and 17 ' C. during the various tests .
The fire was mounted on a special stand , with its flue outlet under a large hood communicating with a chambered wall and chimney which gave a sufficient draught to prevent any of the products of combustion getting into ; the room .
The five bolometer readings in each test were taken at radial distances of 34*4 inches from the centre of the fire : one central reading and four other readings in positions 60 ' N. , S. , E. , and W. respectively , of the central position .
A mean " distribution factor " of 28*36 , as determined for the said fire , was i used in the calculations .
Finally , the authors desire to thank Messrs. A. Forshaw , M.Sc .
, and S. Farrar , B.Sc. , for their assistance in connection with the experimental part of the investigation .

