Fuly 23, 1885] 
the position of the lines in the spectrum. It is obvious, 
therefore, that during an exposure of any considerable 
duration, such as is often necessary with faint spectra, 
this change of position of the lines due to temperature- 
change would absolutely destroy the definition n othe 
photographic plate. The shifting of the lines on the 
plate were found to amount to o'r inch for 1° F. An 
even-temperature box of metal containing cotton was 
made, and the prism arranged within it. The tempera- 
ture was regulated by a thermostat contained within the 
box, consisting of a compound bar of brass and ebonite, 
which turned on or off the gas as necessary. Afterwards 
one even-temperature box was placed within another, and 
it was then found that the temperature could be kept uni- 
form for a long time within or F., and then the lines did 
not shift at allso much as the distance between the sodium 
lines. With this arrangement many hours were taken by 
the box to settle down to a new temperature, so that, if a 
change of over 10° F. is to be made in the temperature 
of the box, it is doubtful whether the spectrum would 
become stationary in less than twenty-four hours. 
The results have a two-fold bearing. In the first place 
they prove that, by the simple expedient of stirring the 
liquid in the prism, all striae may be prevented and the 
definition rendered perfect. The practical value of this 
simple expedient is very considerable. The Thollon 
bisulphide prism, while giving seven-eighths as much 
dispersion as six flint prisms, gives four times the light in 
the entire spectrum and eight times the light in the 
region near G. For photographic purposes, now that the 
definition can be made permanently sharp and the shifting 
of the lines prevented, this prism must replace trains of 
glass prisms, and even gratings, unless these are of large 
size and are used with telescopes of proportional 
aperture. 
In the second place, this investigation has called atten- 
tion in a very marked manner to the change in refracting 
power with change of temperature. This subject has 
already been discussed by several authors, who agree with 
the statement of Arago and Neumann, that for glass the 
law is the reverse of that given for liquids, and that the 
refractive index increases with the temperature. In the 
case of the Thollon prism the refractive index increases 
as the temperature diminishes. As Mendenhall has 
shown that no change takes place in the angle of the 
prism with change of temperature, it follows that the 
observed change of refractive power of the Thollon prism 
is a differential result due to the excess of the index of 
the bisulphide in one direction over that of the glass in 
the other. 
It will ever be a source of regret that Dr. Draper did 
not live to complete the research to which the foregoing 
investigation was preliminary. With his new and admir- 
ably equipped laboratory and with this powerful and 
thoroughly corrected photographic spectroscope at his 
command, no one can doubt that he would have secured 
with it results of the highest value to astronomical, and 
especially to solar, physics. 
PREVENTING COLLISIONS WITH ICEBERGS 
IN A FOG) 
GE recent accident to the steamer Cz/y of Berlin 
emphasises the importance of devising practical 
methods of ascertaining the proximity of icebergs in a 
fog. The precautions adopted by Capt. Laud, though 
they saved the lives of more than 1400 passengers, and 
prevented serious damage to the vessel, did not prevent 
contact with the berg. Even the ‘‘look-outs” were un- 
aware of the proximity of the iceberg until it was actually 
upon them. 
Under these circumstances the method proposed by 
* From Sczence 
NATURE 
273 
Mr. Frank Della Torre, of Baltimore, deserves consider- 
ation. His experiments indicate the possibility of ob- 
taining an echo from an iceberg when in dangerous 
proximity toa ship. Mr. Della Torre believes that even 
an object offering so small a surface as a floating wreck 
may in this way be detected during a fog in time to pre- 
vent collision. However this may be, it is certain that 
his method is worthy of a careful trial at sea, and that 
preliminary experiments, recently made in the presence 
of Prof. Rowland, of Johns Hopkins University, and the 
present writer, have demonstrated the feasibility of pro- 
ducing well-marked echoes from  sailing-vessels and 
steamboats at considerable distances away. 
_ These experiments were made on the River Patapsco, 
near the head of Chesapeake Bay, at a point about seven 
miles from the City of Baltimore. The party proceeded 
down the river in a steam-launch to the selected place, 
where the distance from shore to shore appeared to be 
about three miles. 
The launch was kept so far from land as to prevent the 
possibility of mistaking an echo from the shore for one 
produced by a passing vessel. 
The apparatus employed consisted of a musket, to the 
muzzle of which a speaking-trumpet had been attached. 
This gun was aimed at passing vessels, while blank car- 
tridges were fired. After a longer or shorter time, 
according to the distance of the vessel, an echo was 
returned. 
The ordinary river-steamboats, and schooners with 
large sails, returned perfectly distinct echoes, even when 
apparently about a mile distant. At shorter distances the 
effects were, of course, still more striking. 
In order to test the effects under the most disadvan- 
tageous circumstances, blank cartridges were fired in the 
direction of an approaching tug-boat. The surface pre- 
sented was, of course, much smaller than if the boat had 
presented its broadside to the launch. As the boat ap- 
proached bow on it corresponded to a target somewhere 
about six feet square, presenting a convex surface to the 
impinging sound-wave. Even in this case a feeble echo 
was perceived when the boat was at a considerable dist- 
ance (estimated to be nearly one-quarter of a mile). That 
any echo should have been perceived at all under such 
circumstances was a surprise. The sound was heard 
only by the closest attention, but in the case of larger 
vessels the effects were very distinct and striking. 
Experiments were made which demonstrated the fact 
that the speaking-trumpet attached to the gun was of 
material assistance in giving direction to the sound- 
impulse, and in intensifying the audible effect. 
Mr. Della Torre claims that a steam-whistle or siren, 
combined with a projecting apparatus like a speaking- 
trumpet, will prove as efficient as the gun. 
During the experiments on the Patapsco River a curious 
rumbling effect like the rolling of thunder was often ob- 
served, which continued for some seconds. A similar 
sound was also noticed, as an echo from a well-wooded 
shore ; but the effect alluded to above could not have 
been due in any way to the land, as the sound commenced 
immediately upon the firing of the gun, whereas the 
shore was distant at least a mile or a mile and a 
half. 
The sound was probably due to the presence of ripples 
on the surface of the water, as the effect was much less 
marked when the surface was smooth. Such a sound 
might prove a disturbing element of importance in a 
rough sea, but would hardly be sufficient to prevent the 
detection of an echo from a large iceberg. Had shots 
been fired periodically from the bow of the Czty of Berlin 
it can hardly be doubted that the presence of an obstacle 
ahead would have been discovered in time to prevent the 
collision that actually occurred. 
ALEXANDER GRAHAM BELL 
