377 On Prof. Jolys Method of Avoiding Collision at Sea .
By H. C. Plummer .
( Communicated by Prof. J. Joly , F.R.S. Received February 10 , 1916 .
) The ingenious method proposed by Prof. Joly* for avoiding collision at sea is so simple , both in principle and in practice , so far as can be judged without actual experience , that it may seem superfluous to attempt to introduce still greater simplicity .
But this quality is so important in expedients of the kind that an attempt may find excuse .
Since the problem of collision is essentially a question of relative position , it is naturally reduced to the lowest terms of simplicity by employing the relative speed of the approaching ship B. By the familiar artifice of superposing the reversed speed of ship A on both ships , A is brought theoretically to rest without altering the conditions of the problem .
It is therefore necessary that the relative speed should be known .
The method presupposes that the speed and course of the approaching ship B is communicated beforehand to A by wireless .
By a combination of three scales similar to Prof. Joly 's collision predictor , f a seaman on the ship A is in a position to find this speed very simply without any numerical calculation .
He lays the scale a ( fig. 1 ) in a direction representing his own course , and the scale b in the direction representing the course of the approaching a X * P. 176 , supra .
t P. 252 , supra .
Mr. H. C. Plummer .
On Prof. Jolys ship B. The corresponding speeds being OX , OY , a scale c being brought into the position XY measures the relative speed and direction of approach of the ship B. So long as the course of A is unaltered , the point X is fixed , and for any passing ship it is only necessary to rotate b and c in a suitable manner .
If millimetre scales be used , knots may be represented by centimetres , and ample accuracy will be obtained without excessively long scales .
The most convenient form of this simple apparatus is a matter of detail .
The relative speed and course of every passing ship should be recorded .
When the relative speed of the approaching ship B is known , the conditions are considerably simplified .
When the actual courses are used , the locus of B is a series of circles of which the centres advance with A at the time when the successive signals are received .
With the relative speed , A is brought to rest and the locus of B at the successive signals is a series of concentric circles .
The principle involved is now obvious .
The rate of approach to the centre from an outer to an inner circle is a maximum along a radius .
If ( fig. 2 ) the relative course is radial , the change of distance indicated by two signals , at an interval of 2 min. , is equal to the displacement produced by the relative speed in 2 min. If there is perfect accuracy a collision will then ensue .
If the distance indicated by the signals is less , the approaching ship will pass on the one side or the other .
The distance is calculable , but there is nothing to decide between the two courses c\ and c-2 .
The distance indicated by the signals cannot be greater than that which is covered in 2 min. at the relative speed , unless a mistake has been made .
It remains to consider the simplest and most rapid way in which the comparison is to be effected between the actual displacements shown by the signals and the displacements which correspond to the relative speed maintained during successive intervals between the signals .
Prof. Joly has described one method .
Two others may be suggested .
In the first ( fig. 3 ) a proportional scale is used .
The top line is divided to show knots , corresponding to the relative speed .
A vertical line under 30 knots is divided to show miles on a convenient scale .
These divisions are joined to the zero of the horizontal scale by straight lines which correspond to the intervals of time , 2 min. , 4 min. , . . .
Thus , if a vertical scale S , divided in the same scale of miles , be placed at any point of the top line , say at 15 knots , the position of the zero at the bottom of the scale will indicate the time necessary for a given displacement at 15 knots .
Hence it is only necessary to place the scales at the position on the knot line corresponding to the relative speed .
When the first signal gives the distance of the approaching ship , the scale S is moved vertically to this distance and clamped .
Then the position of the bottom of this scale gives at once the Method of Avoiding Collision at Sea .
time at which the collision is to be expected , if it occurs at all .
When the second signal is received the distance is compared directly with the reading of S on the 2 min. line .
If it agrees within the limit of error , danger is indicated .
The distance shown by the third signal is compared with the reading of S on the 4 min. line , and so on .
If several signals can thus be compared with a scale always in view , the advantage will be gained of being able to discriminate to some extent between accidental errors and the cumulative deviations in the readings which indicate a real margin of safety .
For simplicity of explanation the scale S has been taken to be divided so as to show nautical miles .
But the signals give directly a lag in seconds , from which the distance in miles is deduced by dividing by an appropriate factor , 4*3 for aerial and submarine sounds , 5*5 for aerial and simultaneous wireless signals .
This division , though it would naturally be done by a Table , can be avoided altogether .
The scale S can be divided so as to read directly the number of seconds in the time lag instead of the corresponding distance in miles .
No calculation is then involved .
The same object could be secured in another way , if it were desired to retain miles , and there is 380 On Prof. Jolys Method of Avoiding Collision at Sea .
an obvious advantage in this .
Prof. Joly proposes to use a stop-watch to measure the lag .
It would be easy to provide a dial which would indicate miles directly , either on one system of signals , or if necessary on both .
The other method avoids the use of the proportional scale and the scale S. This comes very near Prof. Joly 's own method .
Advantage has already been taken of his Collision Predictor .
It has been supposed that the three scales ay b , and c ( fig. 1 ) are divided so as to show knots .
Instead , let c be divided so as to show the number of seconds in the signal lag which corresponds to the displacement produced by the relative speed maintained for 2 min. , the interval between successive signals .
As a simple numerical process is now intended , an example will serve the purpose of explanation best .
Let the reading on scale c of the Collision Predictor be 3*6 sec. , which corresponds to a relative speed of about 20 knots .
Let the first signal noted show a lag of 18'4 sec. Then the following numbers may be written down :\#151 ; c. Signal .
c. 0 .
0 1 o sec. min. sec. sec. 3*6 0 18 4 18 *4 \#151 ; 2 14 -8 14 *9 + 0*1 4 11 *2 11 *5 + 0*3 6 7*6 8*2 + 0*6 8 4-0 5*4 + 1*4 10 0*4 4*0 As soon as the number 18*4 has been received , the column headed C can be written down at once by subtracting 3'6 in succession .
This shows that a collision will take place , if at all , after the lapse of 10 min. The lag given by each signal is recorded under 0 as received .
The column 0 \#151 ; C records the difference .
Here the second signal gives a distinct warning .
This becomes less insistent with the third signal , but a margin of error must be allowed .
The fourth signal is fairly reassuring , but the fifth signal is .
decisive .
In this example the ships pass within a mile of each other .
The process is the numerical counterpart of the scales used in the first method .
In order that the principle involved may be used effectively , it is clear that the ship which under the rules for avoiding collision at sea is not responsible shall hold its course unchanged regardless of consequences .
Geometrically , the principle may be expressed in the obvious form that the shortest distance between two concentric circles is along a radius .
To this simple fact is due the unique solution of the collision problem from measures of mutual distance .
It is also evident that this use of a minimum The Ignition of Gases hy Impulsive Electrical Discharge .
381 property does entail a corresponding disadvantage .
It implies a lack of sensitiveness , which is geometrically obvious , in the method when the relative course lies fairly near the critical course .
Whether this would give rise to false alarms which might tend to discredit the method , is a question which must be decided by practical experience .
The Ignition of Gases hy Impulsive Electrical Discharge .
By Prof. W. M. Thoenton , D.Sc .
, D.Eng .
, Armstrong College , Newcastle-upon-Tyne .
( Communicated by the Hon. Sir Charles Parsons , F.R.S. Received November 15 , 1915 .
) 1 .
Sparking Distance .
Disruptive electrical discharge through a gas depends upon the number and nature of the molecules between the sparking points.* On the theory log */ p *-/ 3 of ionisation by collision , the sparking distance S where is the number of pairs of ions generated by a negative ion in moving through a centimetre of the gas , / 3 the number generated by a positive ion.f In a mixture of gases , a and f3 both vary with the degree of mixture .
The addition of air to pure helium , for example , lowers / 3 more than a , and increases S. The electric strength of solid and liquid insulators has been shown to have at first the law S = av + bv2 up to values of S measured in centimetres ; following this , a final stage v = A + BS is generally found .
In gases , the measurements of de la Eve and Muller have shown that the first law holds for airJ up to 11,000 volts ; many subsequent measurements can be represented by the second , especially when the gas pressure is reduced to a few millimetres of mercury .
There appear to be no values on record of the change of the electric strength of gases when the pressure is impulsive .
A few of these are given in fig. 1 ; both stages are well marked .
When the mixture is inflammable , the passage of a spark may cause * W. de la Rue and H. W. Muller , 4 Phil. Trans. , ' vol. 171 , p. 102 ( 1880 ) .
t J. S. Townsend , 'Electricity in Gases , ' p. 322 .
J C. P. Steinmetz , " On the Disruptive Strength of Dielectrics , " 'Trans .
Amer .
I.E.E. , ' vol. 10 , p. 64 ( 1893 ) .

