Juty 27, 1899] 
NATURE 303 
coming too large to occupy the end only of the positive carbon, 
and therefore extending up tts side. 
Simple as is this explanation of a very complicated series of 
phenomena, it is the true one; but before proceeding to demon- 
strate its truth it will be interesting to see how the laws for 
the largest silent currents with normal arcs, which have already 
been obtained from the electrical measurements on pp. 283, 
284, may be deduced on the above hypothesis from Figs. 6 
and 7. 
In Fig. 6 (p. 285) we have a series of four normal arcs of the 
same length burning between solid carbons of the same diameter ; 
but in (a) the current is 6 amperes, in (4) 12, in (c) 20, and in 
(d) 30 amperes. The bluntness of the tip of the positive carbon 
may be measured by the obtuseness of the angle ABC. In (a) 
the tip is very blunt, and the area of the crater is certainly less 
than any but its smallest cross-section ; therefore the arc is cer- 
tainly silent. In (4) the tip is less blunt, but the arc is still 
evidently silent ; in (c) the angle ABC is much more nearly a 
right angle, and it is plain that a very small increase in the 
area of the crater would cause it to burn up the side of the 
tip; therefore the arc is near the hissing point. In (d) the 
angle ABC is practically a right angle, the tip of the positive 
carbon is cylindrical, and the crater has evidently burnt partly 
up its side. Thus with a normal arc, keeping the length of 
the are constant and gradually increasing the current, must bring 
us to a hissing point. 
This brings me to the reason for the great importance of dis- 
tinguishing between arcs that are normal and those that are not. 
For although, with normal arcs of any given length, hissing only 
starts when the current is greater than it can be with any silent 
arc of the same length, with a 7zo7-normal arc of 2 mm. I have 
been able to produce hissing with a current of 11 amperes, and 
to have a silent arc burning with a current of 28 amperes, the 
same carbons being used in each case. 
The reason of this is obvious. When the arc is normal, the 
carbon ends and crater have a perfectly definite size and shape 
corresponding with each current and length of arc, and changes 
in these are made slowly, so as to allow time for the carbons to 
assume their proper form in each case. If, however, the current 
be suddenly much increased, when, say, the carbons have 
previously been very pointed, then the area of the crater may 
increase so rapidly that it will extend up the side of the carbon 
and cause hissing, even although the carbons would have shaped 
themselves so that there would have been room for the crater to 
remain at the end of the carbon if the change had been made 
more gradually. 
Suppose, for instance, the end of the positive carbon were filed 
to a long fine point, then a very small current would make a 
crater large enough to extend up the side of the point, and pro- 
duce a hissing arc. If, on the contrary, the end were filed flat, 
so as to have as large a cross-section as possible, quite a con- 
siderable current could flow silently even with a short arc, for 
in that case it would require the current to be very great for the 
crater to be large enough to fill up the whole of the end of the 
positive carbon. 
Next, I have shown elsewhere (Zhe Zéectrictan, 1895, vol. 
Xxxiv. p. 614) that, with a constant current, the end of the positive 
carbon becornes rounder and blunter, and occupies a larger 
portion of the entire cross-section of the carbon rod, the more 
the carbons are separated. Hence the longer the arc the 
greater must be the area of the crater, and consequently the 
greater must be the current, before the crater extends up the 
side of the positive carbon. Consequently, the longer the arc 
the greater is the largest silent current. 
Thirdly, it follows that when the current and the length of 
the arc have been increased to such an extent that the round, 
blunt tip of the positive carbon occupies the whole cross-section 
of the carbon rod itself, no further increase in the size of the 
crater is possible without a part of it extending up the side of 
the carbon. Hence the largest silent current for a positive 
carbon of a particular diameter cannot exceed a particular value, 
however long the arc may be made. And lastly, similar reason- 
ing, used in conjunction with Fig. 7, tells us that the thicker the 
positive carbon the greater must be the largest current that can 
How silently with a particular length of arc, which was one of 
the results deduced from the curves in Figs. 2 and 3. 
Thus the fact that hissing occurs when the crater covers more 
than the end surface of the positive carbon and extends up its 
side, combined with our knowledge of the way in which the 
positive carbon shapes itself in practice, is sufficient to enable 
NO. 1552, VOL. 60] 
us to deduce a// the laws given on pp. 283, 284 which govern 
the largest current that will flow silently with the xormadZ arc 
under given conditions. 
We come now to the question, why should the arc hiss when 
the crater burns up the side of the positive carbon—what hap- 
pens then that has not happened previously ? 
In pondering over this question, the possibility occurred to 
me that as long as the crater occupied only the ed surface of 
the positive carbon it might be protected from direct contact 
with the air by the carbon vapour surrounding it, but that, 
when the crater overlapped the side, the air could penetrate to 
it immediately, thus causing a part at least of its surface to dzr7 
instead of volatilising. Many circumstances at once seemed to 
combine to show that this was the true explanation. The 
dancing circles I observed, and Mr. Trotter’s stroboscopic 
images, how were they caused but by draughts getting into the 
arc? Then the humming noise, which sounds like the wind 
blowing through a crack, was not this probably caused by the 
air rushing through a slight breach in the crater, already getting 
near to the critical size? This air pouring in faster and faster 
as the breach widened would cause the arc to rotate faster and 
faster, sometimes in one direction, sometimes in another, 
according as the draught was blown from one side or the other. 
Then, finally, the air would actually reach the crater, burn in 
contact with it, and the P.D. would fall and the are would hiss, 
In the open arc, whether silent or hissing, the outer envelope 
of the vaporous portion is always bright green. With the hiss- 
ing arc the light issuing from the cra/er is also bright green or 
greenish blue. What so likely as that the two green lights 
should have a common origin, viz. the combination of carbon 
with air? For the outer green light is seen just at the junction 
of the carbons and carbon vapour with the air, and the inner 
one only appears when air can get direct to the crater. 
Again, why does the arc always hiss when it is first struck ? 
Is it not because a certain amount of air must always cling to 
both carbons when they are cold, so that when the crater is first 
made its surface must combine with this air? 
The cloud that draws in round the crater when hissing begins 
would be a dulness caused by the air cooling the part of the 
crater with which it first came into contact, the bright spots 
being at the part where the crater and air were actually burn- 
ing together. In fact everything seemed to point to the direct 
contact of crater and air as being the cause of hissing and its 
attendant phenomena. 
One easy and obvious method of testing this theory immedi- 
ately presented itself. If air were the cause, exclude the air, 
and there would be no sudden diminution of the P.D, between 
the carbons, however great a current might be used. Accordingly 
I tried maintaining arcs of different lengths in an enclosed 
vessel, and increasing the current up to some 40 amperes. Vo 
sudden diminution of the P.D. could be observed with any of 
the currents or lengths of arc employed, although when the 
same carbons were used to produce ofen arcs, the sudden 
diminution of 10 volts in the P. D. between the carbons occurred 
with a current as low as 14 amperes for a I mm. arc. 
It was, of course, impossible, in these experiments, to avail 
myself of an ordinary enclosed arc lamp, such as is used for 
street lighting, since a current of only some 5 or 8 amperes is 
all that is used with such a lamp, whereas to test my theory it 
was necessary to employ currents up to 40 amperes. Accord- 
ingly I constructed little electric furnaces of different kinds, 
one of which is shown in Fig. 8. 
Some curves connecting the P.D. between the carbons with 
the current when the are was completely enclosed in the crucible 
(Fig. 8) are given in Fig. 9. The carbons were similar to those 
used with the open arc experiments (Fig. 1, p. 282), being solid, 
the positive 1 mm. and the negative 9 mm. indiameter. As this 
crucible—the first one made—had no window, the length of the 
arc could not be kept quite constant, but the distance by which 
the carbons were separated was noted at the beginning of the 
experiment, and they were then allowed to burn away, without 
being moved, till the end, when the distance the positive 
carbon had to be moved in order to bring it tightly against 
the negative was noted. Measured in this way, the length of 
the arc was 1°5 mm. at the beginning and 2 mm. at the end 
of the experiment. The current was started at 6 amperes, 
and gradually increased to 39 amperes; then as gradually 
diminished to 6 amperes again, increased to 36 amperes, and 
diminished to 5 amperes, when the are was extinguished. The 
