III .
" On the means of increasing the Quantity of Electricity given by Induction-Machines .
" By the Rev. T. ROMiNEY ROBINSON , D.D. Received May 10 , 1866 .
Among the remarkable results obtained by studying the spectra of electric discharges , is the change exhibited by certain substances when the nature of the discharge is varied .
In general the mere spark shows fewer and fainter lines than when a Leyden jar is in connexion , though the amount of electricity supplied by the machine is the same .
In the latter case , however , the discharge passes almost instantaneously , and therefore its concentrated action will be more powerful .
But , as far as I know , much has not been attempted towards increasing the power of the jar : this cannot be done by increasing its surface ( unless indeed that be too small to condense all the electricity supplied ) ; the supply itself must be increased .
This may be done in three ways : First , the power of the exciting battery may be increased .
This , however , is limited by the risk of destroying the acting surfaces of the rheotome ; and by the decreasing rate at which the magnetism of the iron core increases with the primary current .
In some investigations on the electromagnet ( Trans. Irish Academy , vol. xxiii .
p. 529 ) I have shown that its lifting power L is approximately given by the equation A , B+ ' in which 'I* is the product of the current and number of spires , A the maximum lift of the magnet , and B the T which would excite it to half A. The rate of change is therefore inversely as ( B+t'i)2 .
The results obtained with two of the magnets which I used will illustrate this .
Their A 's are 781 lbs. and 278 .
The first 1000 of ' make their lifts 576 and 235 ; the second 1000 adds to these 87 and 19 ; the third 35 and 8 ; and the fourth only 19 and 3 .
With a primary of 180 spires , T=4000 implies a current which can evolve in a voltameter 34'7 cubic inches of gases per minute , and of course has great de'flagrating power .
There is therefore not much to be gained in this direction .
Secondly , the secondary helix may be made of longer or thicker wire .
It will be shown immediately that the length does not increase the quantity of the current at all , and that the effect of the thickness is limited .
Thirdly , the analogy of the voltaic battery suggests the plan of combining several helices collaterally , as is done when cells are arranged for quantity ; and this I think may avail to a very great extent .
In spectral work , as in most other applications of the inductorium ( as the Germans have named it ) , the breaking current alone is of importance : the other , though equal in quantity , is so much inferior in tension that it is stopped by a thin film of highly rarefied air* .
This current proceeds from two causes .
When the circuit is broken , the current in the primary ceases ; not instantaneouslyt , but in a time which is very small ; according to Edlund less than -L ?
of a second .
During its decline it induces a current in the secondary .
But while it was passing it had magnetized the iron core of the apparatus : this magnetism now passes away , and in doing so it also induces a current in the secondary , which lasts longer and is more powerful than the other .
I compared the two by measuring those given when the core was removed from a primary , and when it was in its place : they were as 1 : 862 when the rheotome made 17 discharges in a second ; so that in round numbers the electric induction was only a tenth of the whole effect .
I shall therefore in what follows confine myself to the magnetic induction .
If y be the magnetism at any time t , M its maximum , P the potential of the magnet on the secondary helix , I the potential of that helix on itself , r the resistance of the secondary circuit , q the secondary current at the time t , we have , as is well known , P die _ do q= .
x d ... ... . .
( a ) r dt rdt the last term being the counter current produced by the reaction of q on the helix .
If the inductive coefficient of electric action be different from that of magnetic , II should be multiplied by a factor e , which in ( b ) will multiply I in the exponent and denominator .
I see no reason why they should differ unless some work be lost by molecular changes in the core when excited .
The great difference between the two currents which I have just mentioned arises most probably from the different values of t in the exponentials .
To integrate ( a ) we must assume some relation between die and dt .
The most probable is that the loss of magnetism is as the magnetism , which gives dy= -dt , whence y= Me- " .
Putting b = , and supposing q to vanish with t , PMX { ( } ... .
*)r)ir The total current in the time t=-= dt s--dy-Id , and hence putting F for -P ( M-y ) and c for ~i , we obtain rm The expression for the current due to the electric induction will be similar to this , so far as having the factor F and the exponential .
If , as is not unlikely , the relation of dE to dt be the same as for the magnetism , they would only differ in the values of p and c. The quantity F is the current which would be produced were it not for the inductive reaction of the current on itself .
It is as P directly and r inversely .
The first of these , P , is as n the number of spires in the helix* multiplied by a rather complex function of its length and diameter , which is constant when they are given .
The second , r , may be assumed proportional to the length and section of the wire of the helix , for in general the other resistances of the circuit are comparatively small .
Hence it follows that If two equipotential helices equally excited be placed in series , the tension will be doubled ; but the current will be intermediate between that of each , for F= 2P , and if r=rw it will not be changed .
If they be used collaterally ( their homonymous terminals connected ) , P remains unchanged , and therefore the currents of the helices are simply added. .
If there be an external resistance , allowance must be made for it .
This may be extended to any number of helices ; for calling the external resistance p , we find P Pt PP + ... .
F , T 1 o+ n +. .
s ... a ( .d 1 Pr ?
t..9.+ r , j The constant c must be a small fraction , for in any ordinary work of the inductorium the residual ' magnetism of the core is very feeble .
As i t=e-~t , pt must be large ; and as t for wire cores does not exceed a few hundredths of a secondt , , must be very large .
Rev. T. 1I .
Robinson on increasing the The potential II is as n2 multiplied by another function of the length and diameters of the helix ( see Maxwell 's valuable paper " On the Electromagnetic Field , " Phil. Trans. 1865 ) ; and the term b is always less than unity .
When helices are consecutive their rI 's are added , not when collateral .
From this it follows that --c may be neglected ; that -1is nearly unity ; and that the difference between F and ( D increases as 6 diminishes .
When equal helices are consecutive , 6 as well as F are unchanged ; therefore so is 4 .
When they are collateral , each separate b remains unchanged ( unless they be so close that they react on each other ) ; and therefore , as with F , the resultant ?
is the sum of its components .
If the resistance of the wire be diminished by increasing its section without making much change in the dimensions of the helix , b is diminished , and therefore the coefficient of F. It is evident from the form of equation ( c ) that ?
has a maximum for r , and that beyond this there is actual loss of power in increasing the thickness of the wire .
It remained to test these views by experiment , but the task has some difficulties .
A single discharge of inductive electricity is usually determined by the swing which it causes in a galvanometer needle ; but it is scarcely possible to get two discharges exactly equal .
The slightest variation in the manner of breaking the circuit , the least oxidation or roughening of the surfaces where the break is made , change the result ; and therefore it seemed best to take the actual working of the inductorium , in hopes that the average of some thousand discharges must be near the real value of the current .
The rheometer which I used is Weber 's ( for the use of which I am indebted to the kindness of Mr. Gassiot ) , and it showed an amount of fluctuation even greater than I expected .
With every precaution as to the action of the rheotome , the mirror of the Weber never becomes stationary , and the oscillations are irregular ; twelve of them were taken for each set , of course read at each end and reduced by the usual formula ; yet the sets differ so much , that I only offer their results as tolerable approximations .
Two facts illustrating this uncertainty may be mentioned .
With a mechanical rheotome driven at a uniform speed , and its acting surfaces platinum , the ratios of the current wereWhen set so that the point rises but little from the anvil. .
1'0000 Rise greater ... ... ... ... ... ... ... ... ... ... ... . .
] 7894 Rise still greater , tension of spring greater ... ... . .
19685 Rise still greater , tension further increased .
1'8371 Here a slight change of the adjustment nearly doubles the action of the inductoriulm .
Another cause of uncertainty is the variable speed of the rheotome .
In general it is worked by the primary current , and therefore is affected by fluctuation of the battery and the extra current of the primary .
The me chanical rheotome which I have mentioned is driven by clockwork , and its speed can be exactly regulated .
With it I obtained 1 .
Time of rheotome 's stroke =0'2865 .
Current =-1'0000 2 .
, , , 0-2020 .
, , 0-9298 3 .
, , , , 01862 .
, , 08741 4 .
, , , 0-1381 .
, , 07235 5 .
, , , 0-1219 .
, , 0-5771 6 .
, , , 0-1201 .
, , 05588 7 .
, , , , 00983 .
, , 03788 8 .
, , , , 00883 .
, , 02823 9 .
, , , , 00671 .
, , 02478 10 .
, , , , 0-0476 .
, , 02007 11 .
, , , , 0-0278 .
, , 01946 12 .
, , , , 00196 .
, 0-1273 13 .
, , , , 00098 .
, , 00470 No. 10 was taken with a mercurial rheotome ; and the remaining three with a spring one , such as Mr. Lad applies to his inductoria* .
This decrease of power is due to the core requiring time to be mag .
netized .
Suppose the current =A+ Bt A being that caused by the electric induction , B by the magnetic , I get from the above by minimum B squares A=0'0802 ; B-=4'1713 ; =508 ; which values represent the observations pretty fairly , the probable error being + 0'049 1 .
Three cells were used here : on another trial with five I had -= 129 , confirming a previous remark that the electric induction increases faster than the magnetic .
Hence much power is lost by working at too high a speed .
The inductorium which I use consists of a strong oak table , on which are fixed vertically four primary helices , their axes being 12 and 18 inches apart ; at which distance the mutual action of the secondary helices is scarcely sensible in the Weber .
I denote these primaries by P ' , Pt ' , &amp;c. Their wire is No. 12 ; Pt and Pt !
are 12-5 inches long ; they have , in four layers , the first 383 spires , the second 343t .
Their cores of iron wire , No. 18 , are 1'12 diameter .
PlT and Piv are 13-5 inches long ; they have each 181 spires in two layers , and their cores are 1'60 diameter .
They are all insulated by strong glass jars , and their connectors are so arranged that the current can be sent through any one separately , or through all at once .
The normal arrangement is that the battery-current passes through P1p , then through Pf and P'T collateral , and lastly through P'v .
Thus the 4t or exciting power of each primary is nearly the same .
On a shelf below stands a Fizeau 's condenser , each of whose coatings is 120 feet divided into k For the first nine of these the time was given by the clockwork of the rheotome ; for the rest by dropping sparks on a slip of prepared paper , which was made to travel at the rate of 12 inches per second .
t The difference arose from the cotton lapping being thicker in P " .
five sections .
This , though so potent in respect of sparks , does not affect the quantity , which with it I found as 1'0000 , without it 0-9948 , a difference not worth noting .
The case , however , would be different if there were a gaseous interval in the circuit * .
Beside this shelf is a bracket which supports rheotomes of various kinds .
Over the jars can be put any of the secondary helices , the constants of which are given in the following Table : The potential P and resistance of the first one , which I take as a standard , are assumed = 1 .
TABLE I. Feet of Diameter Entire ghtPotential .
lsistance .
wire .
of wire .
ayer .
Spire .
diameter .
HeighResistance .
in .
in .
in .
G ... ... . .
17,070 0-0092 73 I3,655 6'84 40 00000ooooo 0000oooo H ... ... .
17,070 0-0092 73 I3,655 6-84 40 0000ooooo 085270 I ... ... ... 8 , 0'I53 55 6,570 6'72 4'0 0'481I4 ooo0079 K ... ... ... 8,0 o'oI53 55 6,570 6'72 4'0 0'48I14 100oo79 M ... ... .
8,130 0o'o92 29 6,524 6'2I 6'o 0'47520 0'08073 N ... ... . .
8 , I30 ' ooi92 29 6,524 6*56 5'9 0'47520 o'08z82 A ... 7,000 0-0107 ... 6 , i89 5'93 3'5 044673.8o B ... .
, ... .
9,200 o'o107 ... 8,135 5'93 4-0 0o60272 J50 The first six were made by Mr. Lad , who also determined for me the length and number of layers .
The thickness of the wires was measured by me with a fiihlhebel which read to 0'0001 : each is a mean of ten measures at different places , for no wire that I have ever tried is quite uniform .
The two last are experimental , their wire not being lapped , but merely insulated by a varnish of wax and rosin , as proposed by the late Dr. Callan : this plan does well for quantity , but cannot be trusted for any high tension .
The potentials were computed , supposing the helices at the middle of tke primary Pht ( where they are a maximum ) .
For GI computed them in four other positions , and had the curiosity also to measure the currents .
Distance of G from centre =0 , potential 1'0000 , current 1 0000 , , , , 1 , , , 0'9842 , , , 0'9856 2 , , , 0'9790 , , , 0-9746 , , , , 3 , , , 0'9488 , , , 09488 , , , , 4 , , , 0-8798 , , 0'9181 All but the last agree tolerably .
For positions of M and N , which were not central , they were specially computed .
The resistances were obtained by including in the circuit of a small Grove 's cell and a tangent rheometer of 950 spires , first the helix G , then that to be determined , and lastly the sum of it and G. Assuming G= 1 , we have three equations to determine the r of the helix in question , the remaining r of the circuit , and the electromotive force of the cell .
In deducing the currents , the term involving sin20 was used with a coefficient obtained by integrating through the length and breadth of the rheometer 's coil ; and as its mean diameter is 6*4 times the length of its needle , and 0 never passed 54 ?
, I think the numbers of the Table are true to the last decimal .
For brevity I symbolize the combinations , when in series , as a sum , G+ I ; when collateral , as a product G. H ; and for a reason which will soon appear , when two are on the same primary I denote them by a new letter , as V=G+I .
At first I used I and G on Pt ; K and H on p"v .
To prevent disruptive discharge , they were kept 1*5 inch asunder by disks of baked wood .
The results are given in TABLE II .
Sum of Name .
&lt; .
FB.'* Spark .
Sets .
c ntb .
F components .
in .
G ... , ... ... . .
oooo 00oooo 3'95 ... I00000 oo ... ... 'o 00779 ... ... ... . .
I'000ooo 3'37 ... . .
I00000o 0'985385 G+H ... ... 1I0348 I-0096+. .
3 102497 ... 2'00000 0-92306 G.H. .
, ... -9988 2'0223 ... 4 98837 2-0855 I'0oo I ... ... ... ... 2-5343 4'3635 1'3 I 0'59x67 .
'z23149 0'46542 I+K ... ... 27654 4'5358 2V49 3 o'60969 0 ... .
o-46298 0'44872 ( LI-K). .
3'552 6'5756 o'8o 4 0'46462 46220 V+V ' ... ... 0'9996 1'3776 9'35 3 0'72562 ... ... 2z46298 o083075 V.V ' ... ... '9434 2'7319 4'39 I 0'71138 F is computed with a correction for the place of the helix on the primary , and with a resistance which includes that of the Weber =0'00779 .
The Weber must have a considerable n of its own ; but as I did not know its constants , and as this r must vary with the deflection , I did not compute it .
Possibly some of the discrepancies may be owing to this .
The sparks show the difference of tension ; they were taken with platinum points , and when the machine was excited by four Groves .
The column , sum of components , gives for the collateral combinations the values which arise from adding the 4F of each helix , taking into account the Weber 's resistance .
1 .
It will be observed that G+H with twice as many spires , and V+ Vt with thrice as many as G , give the same current ; so also that of I+ K is near that of I. 2 .
On the other hand , G. IH is twice G , and V. V ' twice V+Vt .
3 .
It is also manifest that the effect of I is not proportional to its diminished resistance : its F is 4'4 times greater than that of G , but its actual current is only 2'5 .
This is at once explained by its b being so much less .
So also the ratio of the theoretic to the effective current is nearly unity in G+I , while it is only 0'73 in V+Vt , and for the same reason .
In G. H. ( I+K ) the ratio is still smaller ; but I shall recur to this .
I now used the four primaries I+ K-L on Pv , single helices on the others .
I had some doubt whether the difference of the cores might not influence the results , and tried this with P ' , P't , and one Pv , whose core =1'90 inch in diameter and 15'5 inches long .
It has 349 spires of No. 15 in two layers .
G was put on each of them , and currents transmitted , which made their ' 's nearly equal .
TABLE III .
Name .
Sect. Sets .
Spark .
'F .
core as P/ ... ... ... ... r'oo r 0414 3'5 2z225 1169 iv ... ... ... ... 204 'o0000 2 3'237 ii68 pv ... ... ... ... .
300 10054 2 ' 2'662 1156 It follows from this that the least of these cores is large enough for the excitation produced by four cells : the size does seem to increase the spark , though this increase may be owing to better insulation of the core-wires in P"t and Pvt .
The following results were obtained with the mechanical rheotome worked at a uniform speed of 13 discharges per second .
TABLE IV .
4 .
F. Sets .
m of .
b. components .
0 00 ... ... 00 ... ... . .
0oooo 00oooo. .
ooooo ... ... I'ooooo ' 00779 L=I--K ... I'8832 4'5o06 19 0'4I752 ... ... 0'6+ 034624G+L ... ... ... I'3149 I'643 2 '-81451 ... . .
16+ 0'75002G .
L ... ... ... 3 '0402 5'3I09 II o'57I59 2'8324 H ... ... ... ... .
4196 I3578 2 I104551 ... ... 'o00000 0'85385 C=A+B ... 1-9882 2'01o2 2 0'99855 ... ... 0'56286+ 0'92704 N ... ... ... ... ... 2-5535 5'4o00 I 0'47202 ... ... o'3 646 0'27756 I ... ... ... ... ... 25580 4'3635 2 0'58623 ... ... 2349 0'46542 K ... ... ... ... ... 26465 4'3635 5 o6o65i. .
o0'23I49 o'46542 G.K ... ... ... .
3-5558 5'2464 I 0'67776 3'8594 - .
I. l T ... ... 4'3I79 6'5756 3 o065666 4'1799 G.L.C ... ... .
5'460 6'9400 3 o'74I53 4'6932 G.L.C.II ... 6'0758 8'zo8i 3 0'74030 6'0488 M I+N ... ... ... 252o7 5'4586 2 0-46179 0 ... . .
o'63292 0'26862 =0 ... ... ... ... I76I6 5'4586 4 0'32271 ... ... 0'73130 0'23248 S=M+N ... I'6687 4'9644 I 0o33614 ... . .
0o658Io 0'28543 L. 0 ... ... . .
3'4596 9'5774 I 0'36124 3'4996 MNI K ... 8-'939 15'7398 0'52058 8'2736 * 2'5 sets taken on the first cday were very unsteady .
The one taken on the second was close , and gave 1'0267 .
t At the same time I tried two helices similar to A and B , except that their wire is varnished iron , which were given to me by Dr. Callan .
Their ~4 is 0-6887 , and their spark only 0'524 in .
The 4 of AB is 2-9138 times as great , a difference caused solely by the greater resistance of the iron .
1 .
As before , two helices in series give no increase of quantity ; M+ N is the same as N , G+L nearly the mean of G and L. 2 .
The quantity of collateral helices is seen in column 6 ( as in Table II .
) to be the sum of the separate actions of each .
The discrepancies are not greater than what can be explained by the uncertainties inherent in these measures , which I have already described .
One apparent exception is a strong confirmation of this.rule the case of G. H ( 1 + K ) .
Its observed ( I=3'0552 , while the sum =4'6220 .
G and I being on the same primary are excited together ; but in measuring either , as there is no current in the other ( but merely a state of tension ) , their n is not changed .
When , however , they are connected collaterally , the currents react on each other , their [ I 's are increased : the V 's are thus diminished , and therefore their resultant is the sum of quantities less than those used in the computation .
The b 's in this case become-for G 0'94289 , for H 0-79885 , and for I+ K 0'34588 , which are quite sufficient to account for the difference .
3 .
As with I in Table II .
, so here it will be observed that N has less relative power than either I or K ; its actual power is even less , though its theoretical force exceeds theirs in the proportion of 5 : 4 .
This is explained by its b being so much smaller ; but it gives this important information , that , at least in helices of these dimensions , nothing is gained by using wire thicker than that of I , or of an inch* .
4 .
The effect of L is far less than that of I+ K. In the first the helices are on the same primary , in the other on separate ones .
In the former case the II is larger , for it is the sum of the [ 1 of each on itself , and those of each on the other ; b therefore is less .
Besides , the potential of the core on the helices is less than when each of them is central on it .
The difference is even more remarkable in O as compared to its elements M+N , its effect being only 0'7 of the other , and 0-3 of the theoretic power .
The same disparity of course prevails in their combinations ; O. L giving only 3'46 , while the same four helices arranged on separate primaries give 8 ' 19 .
The combinations G. H ( I+ K ) and G. L. H have the same helices ; but in the first two were on the same primaries .
As , however , they were 1*5 inch instead of 0'5 apart , the [ I was not so much increased as in the other cases , and therefore there is not quite so great a decrease of power .
The following practical maxims may be deduced from the experiments and reasoning whiph I have related .
The attention of instrument-makers has been chiefly directed towards increasing the length of spark given by these machines , and in this they have succeeded to a surprising degree ; but in doing so they have not added to the quantity of electricity which is produced by them .
This , however , is by far the most important object ; for in most applications of the inductorium all tension above what is necessary to force the necessary quantity of current through the circuit is useless , nay sometimes injurious* .
I am inclined to think that a tension which gives sparks of 4 inches will be found quite sufficient in ordinary cases , and this will be given by about 20,000 spires ; all beyond only adding to the weight of the instrument , its cost , and the difficulty of insuring its insulation , It must be kept in mind that the mere quantity is independent of the length of wire : I actually found it the same for a flat spiral of 21 spires and for a helix of 13,655 .
It is not , I believe , ascertained what is the best proportion of height and diameter for a secondary helix of a given number of spires .
It is generally made as long as its primary , though perhaps not on any definite principle .
The magnetic potential P is in this form a little greater than in that which I used , but so also is II : the length of wire is less , which increases F , but also decreases b ; and a priori it is not easy to decide which way the balance inclines .
The I is something less if the spires be in separate sections than if they be in one continuous coil .
The dimensions of the core do not seem to be of importance as to quantity within the limits which I tried ; their length seems to increase the tension .
The quantity is greatly diminished when the rheotome works rapidly ; and in spectral work the probable limit of its slowness is that the impression on the eye shall be continuous .
The quantity increases with the diameter of the wire up to a maximum , which is attained when this is about the sixty-fifth of an inch .
Helices may be combined either for tension or quantity without much loss of these respective powerst .
If for the first , they are combined in series ; the general tension is the sum of the individual ones , and in this way we can obtain sparks of a length limited only by the strength of the insulator which is interposed between the primary and secondary helices .
If the latter be all of the same wire , the quantity remains unchanged ; if they differ in this respect , it will be intermediate between the weakest and strongest .
If they are combined for quantity , they must be set collaterally , i. e. all their positive terminals connected , and all their negative .
The resulting current will be the sum of all the separate ones , but the tension is not increased ; the sparks seem even a few hundredths of an inch shorter , but are much denser , and in the higher combinations approach to the character of a jar discharge .
Hence there is no risk to the apparatus by extending this mode of combination to any extent .
It deserves notice that the helices need not be equal in tension or resistance ; thus the arrangement G. K gives little less than the sum of its : components , though K has only half as many spires as G and but a tenth of its resistance .
In combining these instruments , the primaries should not be consecutive if of large numbers , for so the action of their extra-current would be very destructive to the rheotome ; with P'+P"t containing 726 spires in series the spark in the mercurial one is almost explosive , but when they are collateral it works quietly .
Were , however , ten or twelve to be so combined , it would require a battery of very large cells to maintain the current , and it is better to have a separate battery for each pair of primaries .
In this I find no difficulty ; the negative* poles of all the batteries are connected with the mercury of the rheotome ; from its platinum point , separate wires go to the entering bind-screw of each primary , other wires go from their exit bind-screws to the positive poles of their respective batteries , and thus their action is perfectly simultaneous .
Of course , if many batteries were used , the current in the rheotome might be too powerful , but then there would be no difficulty in having separate rheotomes worked by one electromagnet , and ( at least with the mercurial form ) adjusting them by a revolving mirror to perfect synchronism .
In this way I feel sure that we can attain an amount of electric power which has not yet been approached by the inductorium , and which may be expected to be a most powerful means of research in those inquiries to which I referred at the commencement of this paper .
At the head of these stands the palmary discovery of Mr. Huggins , that there are nebulae and comets whose matter possesses spectral attributes not corresponding to that of the sun , the stars , or our own earthly elements .
Is that difference an indication of some body sui generis , or a mere result of peculiar temperature or other molecular conditions ?
Is , for instance , the bright line , corresponding to one of nitrogen , which occurs , we may say , normally , produced by nitrogen as such ?
If so , what has blotted out the other bright lines of that magnificent spectrum ?
Is it due to an element of nitrogen , dissociated by some enormous temperature from other elements , perhaps from hydrogen , one line of which is also present ?
And the third line , elsewhere unknown-is it the herald of a new body , or merely a derip vative from another spectrum ?
We cannot even hope for an answer to these questions till the spectra of at least those elements which seem cosmical have been examined through a range of temperature extending from the lowest that developes in them luminous lines , to the highest that is excited by the most potent electric discharges which we can produce and control .
Now , to obtain such a graduated range , the plan of combination which I have been describing seems well fitted .
It , of course , cannot be expected to equal , under any extension , the wonderful voltaic battery of Mr. Gassiot ( at least its arc-discharge ) ; but how few can avail themselves X If the mercury be made positive , each discharge makes a sharp report and blows about the metal and alcohol in most unpleasant profusion .
of such an instrument as that !
But if , as seems probable , we can without much difficulty increase the heating-power of the induction-discharge an hundredfold , we shall have made a very great step , and by means which are everywhere accessible .
Inductoria are common ; there are few situations where a physicist cannot obtain access to several , and combine them as Despretz did the voltaic batteries of Paris to make the experiments which have thrown such splendour on his name .

