191 Further Experiments on the Variation of Wave-length of the Oscillations Generated by an Ionic Valve due to Changes in Filament Current .
By J. H. Vincent , M.A. , D.Sc .
( Communicated by Prof. W. H. Brag , F.R.S. Received February 19 , 1920 .
) Introduction .
In the course of a research* on the effects produced on the wave-length of the oscillations given out by a triode maintained oscillator when various changes were made in the circuit , Eecles and Vincent showed that in an oscillatory circuit maintained by a B valve with a grid coil coupling ( like the two oscillatory circuits in fig. 1 ) the wave-length varies slightly with the Fig. 1 .
filament current and has a maximum value for a certain filament current .
In all cases investigated it was found possible so to adjust the coupling between the grid coil and main oscillator coil as to allow the filament current to be set to make the wave-length a maximum .
Elimination of Possible Effects of Self-induction in the Filament Circuit .
Experience gained in dealing with oscillatory circuits had emphasized the difficulty and importance of avoiding changes in other variables of a circuit when it is desired to study the effects due to the alteration of some one variable .
In the former work the changes in the filament current have been biought about by adjusting a rheostat in which the motion of the slider altered the portion of a solenoid included in the filament circuit .
The self-* Eccles and Vincent , ' Roy .
Soc. Proc. , ' A , vol. 96 ( 1920 ) .
Dr. J. H. Vincent .
Wave-length of the induction of the filament circuit was thus altered as well as the resistance \ and it was quite conceivable that the change in the inductance might have some effect .
Experiments were carried out in which the variable resistance consisted merely of an electric arc carbon , 27 cm .
long and 05 cm .
in diameter having a resistance of about 1*2 ohms .
One end of this was copper plated and soldered to a wire while the sliding contact consisted of a spring of copper .
The wave-length was again found to have a maximum value for a definite filament current .
In this experiment , however , the self-induction of the filament circuit was not absolutely constant , for the contour of the circuit is appreciably altered by the motion of the slider .
An arc carbon rheostat was then made in which the contour of the circuit is very little varied by the motion of the slider .
The Arc Carbon Rheostat .
This consists of an arc carbon mounted on a vertical side of a horizontal br#ss bar of square cross section .
The carbon is supported mechanically by cementing it to a strip of vulcanised fibre which in turn is cemented to the vertical side of the brass bar .
The carbon may be either mounted unprepared or it may have its resistance very materially increased by carefully grinding and filing .
Before mounting the carbon rod or strip , one end is copper plated and soldered to a wire which is connected to one of the terminals of the rheostat .
The other terminal is in electrical connection with the brass bar and with a horizontal brass rod vertically under it upon which the slider sits .
The spring of the slider forces a blunt metal knife edge against the carbon and the reaction is taken by the other side of the brass bar .
A current entering by one terminal traverses the brass rod and bar to the slider , passes along the rest of the carbon to the copper-plated end and hence to the other terminal of the rheostat .
By the use of this rheostat sufficient variation in current may be obtained to draw a filament current wave-length curve while the variation in the self-induction of the filament current circuit is sensibly constant .
In fig. 2 is shown the small portion of the curve when the filament current varies from about a fiftieth of an ampere below to a fiftieth of an ampere 3 89 I__________1______ 1_______________I-----------J 0-70 071 o 0-73 0-74 Filament current in amperes ( Unconnected readings ) Fig. 2 .
Oscillations Generated by an Ionic Valve .
above the current for which the wave-length is a maximum .
The whole change in the ordinates represents a change in wave-length of 22 cm .
the wave-length being about 3000 metres .
A larger part of the curve is shown in fig. 3 , which was obtained by using a thinner strip of arc carbon in the - 6 SU3 o m n\#163 ; jf / i i i V \ \ \ 0-68 069 070 071 072 0-73 0-74 0-75 0-76 0-77 0-78 0-73 Filament current in amperes ( Uncorrected , readings ) Fig. 3 .
rheostat .
The vertical scale of this figure is the same as in the last one , a degree of the small condenser ( C3 , fig. 1 ) corresponding to a third of a metre in the wave-length .
Use of Different Methods of Linking and Rectifying .
The results given by Eccles and Vincent and those already described above have been obtained by employing the differential wave meter described in the former paper .
In this apparatus the linking circuit consisted of two coils connected merely by leads .
( To avoid circumlocution , a third circuit influenced electro-magnetically by two main oscillatory circuits of a differential wave meter is called a linking circuit .
) Each one of these coils was placed near one of the main oscillator coils while the telephone was inserted in the plate battery circuit of one of the oscillator sets and was shunted with a condenser .
This oscillating circuit had its main condenser shunted by a variable condenser of small capacity ( as C3 , fig. 1 ) the readings of which combined with the constants of the circuits enable small changes in the wave-length to be measured .
There are two objections to this form of linking and rectification .
The first is that the electromagnetic coupling of the two Dr. J. H. Vincent .
Wave-length of the main circuits is liable to introduce trouble .
The effective coupling was reduced by inserting a crystal detector in the linking circuit .
The second objection is that the arrangement was unsymmetrical .
The telephone and its shunting condenser were removed from the plate circuit and the telephone was put in the linking circuit in series with the crystal detector and the two coils .
Curves giving the change of wave-length with filament current were drawn and were found to be of the same character as those previously obtained .
The note in the telephone was not as loud as before , but the intensity increased considerably on shunting the telephone with a condenser .
By using two crystal detectors in series in one of the leads connecting the linking coils , and joining up the telephone to a point between these detectors and the other lead , a sufficiently loud note is obtained without shunting the telephone with a condenser .
The whole differential wave meter , as now arranged , is shown diagrammatically in fig. 1 , in which the two carborundum brass detectors are shown joined up in the same sense in one of the leads of the linking circuit .
It was found that reversing the connections of one of the detectors made no difference .
Whether this method of linking and rectifying is better than that used in the previous work is not certain .
The note in the telephone was not quite as loud as in the original arrangement , unless the linking coils were placed nearer the oscillating coils , so that it may be that the gain in electromagnetic freedom is at the expense of increased electrostatic coupling .
But the point at present is that the arrangement provides a third method of linking and rectifying , which can be used to test the legitimacy of the earlier results .
Curves drawn with this apparatus were similar to those already obtained , so that we may regard the existence of a maximum wave-length for a particular current as a genuine effect , independent of the method by which the change in wavelength is studied .
Filament Current Wave-length Curves of Different Valves .
Several valves , but all of the B type , have been tested .
In all cases it has been found possible to adjust the mutual induction of the grid coil and main oscillator coil , so that the maximum wave-length is given when the filament current has a value which enables the valve to maintain the oscillations .
The actual shape of the curve is a function of the valve and of the circuit in which it works , but the typical curve rises very slowly at first as the filament current is increased , then becomes steeper until the maximum is approached ; the curve ends abruptly on its descending side , when the current in the* filament passes a certain value .
In one valve the initial portion of the curve shows a distinct but slight bend towards the axis of the Oscillations Generated by an Ionic Valve .
current : with this valve a minimum wave-length is obtainable in addition to the usual maximum .
Amplitude and Wave-Length .
It is , of course , well known that , as the filament current is gradually raised from a very small value , the oscillations cannot be detected until a definite filament current is reached , and that the amplitude of the oscillations at first increases as the current rises .
This rise in current is accompanied by an increase in wave-length .
When the current in the filament of a B valve is raised above a certain value , the intensity of the oscillations falls .
We have seen that the wave-length also decreases .
When the conditions are arranged so as to get the maximum wave-length on an accessible part of the filament current axis , the maximum amplitude is probably given by the same current as yields the maximum wave-length .
This last statement is made on the strength of experiments in which the amplitude is assumed to be proportional to the readings of a galvanometer in series with a crystal detector shunting the condenser of a resonating circuit .
Other cases in which the amplitude and wTave-length seem to be correlated are when the resistance of either of the branches R or S ( fig. 1 ) is changed .
Increasing either , lessens the amplitude and decreases the wave-length .
The experiments on this point are not yet published .
We have , then , seven cases , which can be connected together by the following rule :\#151 ; When any one of the changes below is made in a valve set of the type of those in fig. 1 , worked by B valves , the wave-length and amplitude decrease .
1 .
Decrease grid voltage .
2 .
Decrease plate voltage .
3 .
Decrease grid coil coupling with main oscillator coil .
4 .
Increase resistance in condenser branch of oscillator circuit ( R , fig. 1 ) .
5 .
Increase resistance in self-induction branch of oscillator circuit ( S , fig. 1 ) .
6 .
Increase the filament current beyond the value which gives the maximum wave-length .
7 .
Decrease the filament current below this value .
At present , the writer suggests the above rule merely as a useful aid in remembering the sign of the effect on the wave-length of these changes in the circuit .
The effect on the amplitude can in general be deduced by imagining the change pushed to the extreme limit , and then the above rule enables one to find the sign of the consequent change of the wave-length .
No claim is put forward that the curves giving the simultaneous values of the wave-length and amplitude are parallel or have absolutely coincident 196 Mr. T. Y. Baker and Prof. L. N. G. Filon .
Longitudinal maxima , but it seems probable that there is a close connection between them which will be disclosed by further experiments .
The work described above was carried out in Dr. Eccles ' laboratory , under his direction and at his suggestion .
I wish to thank him for having initiated me into the use of the apparatus , and for constant help and advice during the progress of the research .
On a Theory of the Second Order Longitudinal Spherical Aberration for a Symmetrical Optical System .
By T. Y. Baker , Instructor Commander R.FT .
; and L. N. G- .
Filon , M.A. , D.Se .
, F.B.S. , Goldsmid Professor of Applied Mathematics and Mechanics in the University of London .
( Received December 2 , 1919 .
) ( Abstract .
) 1 .
The object of this investigation is to establish a formula for the longitudinal spherical aberration of rays which traverse a symmetrical optical system in an axial plane that shall be capable of fairly easy computation for any combination of lenses , and at the same time shall be accurate to the second order and free from certain important difficulties of convergency which occur in certain neighbourhoods when we attempt to use for the longitudinal aberration the method of aberration of successive orders .
From the point of view of the optical designer , the usual theory of aberrations , which , for all practical purposes , is largely restricted to the first order , is known to give an unsatisfactory approximation .
In practice , the designer adopts a semi-empirical method of tracing a number of rays through the system by means of the trigonometrical equations , a method which is laborious and lengthy , and which can at best give only incomplete information and very limited guidance for effecting improvements .
2 .
The type of formula established in the present paper was originally suggested by an empirical formula discovered by the authors* when trying to fit the aberrations of a series of thick lenses , obtained from the ray paths calculated trigonometrically .
This formula , when tested , fitted the results * ' Transactions of the Optical Society , ' December , 1918 .

