]\gt ; The Magnetic Storm of December 16-17 , 1917 , as Recorded at Eskdalemuir Observatories .
By C. CHREE , Sc. D. , LL. D. , F.B.S. , Superintendent Kew Observatory .
( Received March 19 , 1918 .
) S1 .
In a recent paper , * I described a magnetic storm occurring on August 22 , 1916 , in which the disturbances recorded at Kew and Eskdalemuir ervatories were closely similar i-n type .
The magnetic storm on December 16-17 , 1917 , affords in many ways a remarkable contrast , the changes recorded at the two observatories differing notably in type .
The disturbance was the largest of the year .
In the Kew declination ( D ) and horizontal force ( H ) curves , a prominent feature was a succession of oscillations with periods of about 20 minutes , and it was the hope of these with corresponding oscillations at Eskdalemuir that suggested the present investigation .
Dr. Crichton Mitchell , the Superintendent of Eskdalemuir Observator .
, kindly sent the original curves , so the comparison was made under favourable conditions .
One of the obstacles to the identification of corresponding movements at the two stations is that , while and are recorded at New , the north and west ( W ) components are recorded at Eskdalemuir .
On instituting a minute comparison between the and curves , it was found that , while oscillations with periods not far from 20 minutes were prominent in both , the times of the turning points by no means always agreed .
A turning point in the curve , for instance , might be represented by a short arrest or slight temporary reversal of movement in the curve , the original direction of the movement being almost immediately resumed .
Again , towards the end of an oscillation , the trace be nearly level for some minutes , or there might be a small short period oscillation , so that more than one choice was possible when assigning a time for the end of one major oscillation or the beginning of the next .
Thus , when one to compare Kew with Eskdalemuir curves , and Kew with Eskdalennuir curves , the fact that corresponding movements were difficult to identify is not surprising .
Both observatories record vertical force ) , but the Kew curves are so much disturbed by artificial electrical currents that they cannot be relied on for minor details .
There were also difficulties special to the particular storm .
Some of the and movements at Eskdalemuir were so rapid that the photographic trace was in places almost invisible .
While the time marks 'Roy .
Soc. Proc , vol. 93 , p. 177 .
VOL. XCIV.\mdash ; A. 2 Dr. C. Chree .
occur on the base lines as well as on the curve lines , there is sensible parallax , and when the marks on the curve lines become invisible , ) time cannot be fixed with quite the usual accuracy .
S2 .
The disturbance was preceded by a fairly quiet time , so its commencement may be accepted without hesitation as occurring between 8 and 9 .
on December 16 .
But the subsidence was as usual more gradual .
surements were made of the curves from 8 .
on the 16th to 8 .
on the 17th .
They were taken at four-minute intervals , because four minutes answers very nearly to 1 mm. in the base line .
The curves were not smoothed , and the arithmetic mean of the measurements made throughout each hour G.M.T. was accepted as the hourly mean .
Hourly values were thus obtained from .
to on December 16 , and from .
to .
on December 17 .
The arithmetic mean of these 24 hourly values was accepted as the mean for the " " day\ldquo ; which commenced at 8 .
on December 16 .
Table I.\mdash ; Mean Hourly Values Unit C.G.S. .
The Kew trace records angular changes , but at present a rise of one minute in westerly declination is equivalent to a change of in force The gnetic Storm of December 16-17 , directed south of true west .
The Eskdalemuir trace unfortunately went off the sheet , in the direction of increasing force , several times between .
and the 16th , there no record for some 55 minutes in all , and the longest continuous absence of trace being about 20 minutes .
When the curve was off the sheet , the value answering to the margin of the sheet was accepted .
by the trend of the curve , the margin was not much exceeded in most cases .
But between .
and 18 , when the trace was off the sheet for 20 minutes at a time , the margin may have been considerably exceeded .
An error in one hourly mean implies an error in the mean for the day .
The existence of this uncertainty in the mean value of at Eskdalemuir was one of the reasons which led to the adoption as standard values of quiet day means derived from the previous 24 hours .
Table I shows the departure of the hourly means from these quiet day means , the unit for all the elements 1 C.G.S. netic unit .
Maximum and minimum values are in heavy type .
Every hourly value of at Kew , except that for , falls short of the quiet day mean , i.e. , represents a position of the declination needle to the east of the normal .
The existence of this phenomenon at Kew , while there is no corresponding abriormality in at Eskdalemuir , is one of the most remarkable differences between the two stations .
Another remarkable feature is that while at Eskdalemuir shows the depression customary in that element and in towards the end of a magnetic storm , there is no corresponding sensible depression in at Kew .
V was much enhanced at both places in the afternoon of the 16th , customary phenomenon of magnetic storms .
This elevation of was followed by a rapid fall before midnight , also a customary feat .
ur ; but subsequent to this there was another considerable rise , and a second rapid fall , more especially at Eskdalemuir .
S3 .
Hourly mean values of and at Kew were calculated from the hourly means of and , and the departures of these from the quiet day means are shown in the accompanying diagram , with the corresponding departures in and at Eskdale and in .
The scale of ordinates is the same for the six curves .
Up to 22 .
on December 16 there is a general resemblance between the changes in at the two stations ; but later , especially from 1 .
to 3 .
on the , the changes are more nearly diametrically opposite .
The ranges for the whole 24 hours , as given by the hourly means , were 288 at Eskdalemuir as compared with 279 at Kew .
This gives a very inadequate idea of the excess in the activity at Eskdalemuir .
A better idea is given by the values of the absolute range , i.e. , the excess of the largest over the least value Dr. C. Chree .
recorded at any instanv during tlJe 2 hours .
absolute ranges were 620 at Eskdalemuir in , and 408 at Kew in H. The gnetic Storm of December 16-17 , 1917 .
The resemblance between the diagrams for Kew and Eskdalemuir is not close , but it is not less close than usual between 1 .
and 3 .
on the 17th , when the diagrams from the two stations were opposed .
At Eskdalemuir was markedly enhauced from 13 .
to 19 .
on the 16th , whereas at Kew was markedly } ) essed after 15 .
The ranges from the hourly means were at Eskdalemuir and 183 at Kew .
The difference between the absolute range at Eskdalenluir and the absolute at Kew was even greater , the former being 588 , the latter only 189 The ranges from the hourly means were 446 for Eskdalemuir and for Kew , the absolute being respectively and 259 .
As already explained , the values for Eskdalemuir are necessarily under-estimates .
S4 .
Table II gives the results of measurements of the oscillations of period already referred to .
The different curves were considered and measured quite independently .
A minimum reading was taken as the commencement of each oscillation .
The second minimum for one oscillation served in many cases as the first minimum for the next .
In other cases an interval was left , answering to a portion of the curve which was level or showed one or more short period oscillations , the inclusion of which in either of the two adjacent longer period oscillations would have been arbitrary .
If we started with the assurance that we had to do with a continuous train of oscillations , confused by the superposition of waves of shorter period , we should naturally assign the doubtful piece of curve to one or other of the adjacent longer period oscillations .
If there had been a very pronounced resemblance between Kew and Eskdalemuir curves , and the hypothesis of a small differential time error had markedly improved the coincidence of turning points , it might have been fairly justified .
But , as matters stood , it seemed best to leave the figures exactly as the curve measurements made them .
The result is that 14longer period oscillations were nised in the and curves at Kew , the first 13 of which proceeded with little or no interruption , whilst the 14th followed after an hour 's interval .
The 13th differed somewhat in type from the others .
It possibly represented two oscillations rather than one .
The mean duration for the first 12 oscillations , acceptin the figures in Table II , is in minutes , and in minutes .
If we assume that there was a continuous train of oscillations , and accept the times given for the beginning of the 1st and the end of the 12th oscillation , the mean durations become minutes for and minutes for D. At Eskdalemuir , between 15 .
and , on the 16th , 16 oscillations of longer period were recognised in and 13 in W. Some , however , were of Dr. C. Chree .
The Magnetic Storm December 16-17 , 1917 .
531 decidedly shorter period than the others , and may have answered to portions only of the oscillations recognised at Kew , large short-period oscillations occurring sometimes at Eskdalemuir while the Kew curve showed almost stationary conditions .
The Eskdalemuir oscillations could be brought into fair parallelism with the and oscillations at Kew by assuming that Nos. 3 and 4 at Eskdalemuir together answered to No. 3 at Kew , whilst Nos. 5-11 at Eskdalemuir corresponded respectively to Nos. 4-10 at lCew , Nos. 12 and 13 at Eskdalemuir corresponding to the oscillations similarly numbered at Kew .
We could also bring the earlier oscillations at ] emuir into fair parallelism with the others by regarding the pairs Nos. 3 and 4 , 5 and 6 , 7 and 8 , and 9 and 10 as each but a single oscillation , an swering respectively to Nos. 3 , 4 , 5 , and 6 at Kew .
The Eskdalemuir curves contained some exceedingly large rapid oscillations , practically unrepresented at Kew .
These must presumably have arisen from some source of disturbance which was purely local , or was situated very much nearer to Eskdalemuir than to Kew .
As regards the size of the oscillations recorded in Table II , the mean values of the rise and fall at Kew were respectively 55 and 6 for and and 35 for D. The movements were thus , on the average , decidedly the larger .
If we accept the arithmetic mean of the rise and fall as a measure of the amplitude , we find that the first six oscillations in were very similar in amplitude , while the corresponding oscillations differed widely .
There was , in short , little , if any , parallelism between the amplitudes of the oscillations in the two elements .
If we accept the problematical scheme of correspondence between Kew and Eskdalemuir oscillations given above , we find that the Eskdalemuir ranges were almost invanably the larger , but the numerical results , from a comparison of Kew and Eskdalemuir curves , are too uncertain to be worth recording .
S 5 .
During many of the rises and falls recorded in Table II , the rate of change of the element varied greatly .
Rates of change derived from Table II would thus give an inadequate idea of the rapidity of the more rapid changes .
An independent set of measurements was accordingly made , the results of which appear in Tables III-VII .
They the central time of the movement , i.e. the arithmetic mean of the times of beginning and ending , the number of minutes the movement lasted , and the estimated rapidity of the change of force in 's per minute .
In most cases , where the rises and falls recorded in Table II showed a sudden marked acceleration or retardation , measurements were made on shorter portions of curve having a nearly uniform slope .
In a few cases , however , mean rates are given for Dr. C. Chree .
periods of time , when there was a conspicuous movelnent in one direction , interrupted by only trifling arrests or reversals .
No measurements were attempted of the shorter period movements , mostl small , occupying one or two minutes or less , unless they were quite exceptionally large .
The slope of the curve then affords a ready check on the result derived from the time measurements .
Table III.\mdash ; Bapid Changes in at Kew .
Table \mdash ; Rapid Changes in at Kew .
The Magnetic Storm of December 16-17 , 1917 .
id Changes in at Eskdalemui Table \mdash ; Bapid Changes in at Eskdalemuir .
Dr. C. Chree .
Table \mdash ; Rapid Changes in at Eskdalemuir .
Independent measurements were first made of the absolute times of the beginning and end of each movement .
A differential measurement was then made giving the duration directly .
Finally , the slope of the curve was measured with a special scale .
With times measured only to the nearest minute , no very great accuracy can be claimed for the rates , but the checks mentioned above should be a sufficient protection against large errors .
The probable error is naturally greatest where the time interval is least , and the large rates obtained for some of the shortest movements may thus incur suspicion .
The Kew rates are all much slower than the faster Eskdalemuir rates , and the -fastest of them , per minute in , was derived from an ute interval , and so can hardly be much in error .
The fastest rate for changes at Eskdalemuir , viz. per minute , is based on a 3-minute interval , but that interval was immediately followed by a second 3-minute interval , the change occurring during which had the same estimated rapidity .
The two movements constituted an oscillation in which the to-and-fro movements were practically equal .
The duration of the complete oscillation was thus a particularly convenient one to measure .
If we accept six minutes for it , but suppose that an error made in allotting equal times to the two movements , then the duration of one of them must have been overestimated , and so its rapidity under-estimated .
A somewhat similar argument applies in the case of the next most rapid rate , viz. , 50 per minute .
The most rapid rate for in Table , viz. 50 per minute , is followed by rates of 47 , and 45 per minute , and these four rates arise from intervals which form between them the 17 consecutive minutes 21 .
to .
on the 16th .
The error in the total interval is unlikely to have exceeded one minute , and any redistribution of the time The Magnetic Storm of December 16-17 , 1917 .
amongst the four sub-intervals would almost inevitably have led to at least one rate in excess of any of the four assigned .
The fastest rate for arose from a -nlinute interval .
It was a particularly favourable case for measuring the slope of the curve , as the trace as bold and the slope uniform to the eye .
In the case of .
the longer intervaIs , the assigned rate must in general have been very sensibly exceeded during part of the time .
No movement was included in Table [ , any part of which was lost through the trace going off the sheet .
Several rapid movements were excluded for this reason , but none so rapid as the fastest movement given in the Table .
No rates were calculated for changes at Ksw , because little reliance could have been placed on them .
Table VIII .
Aggregates of apid Changes and Corresponding Mean ates .
Table VIII gives particulars of the aggregates of the rises and falJs included in Tables III-VII .
But for the loss of trace , the time gregate for at Eskdalemuir would have been considerably larger .
While the time regates for the tlifferent elements are not identical we may fairly conclude that during the most active part of the storm changes in and at Eskdalemuir were roughly twice as rapid as changes in , and times as rapid as changes in at Kew .
The changes at Eskdalemuir and the changes at Kew had very similal mean rates , but the fastest changes at Eskdalemuir were decidedly faster than the fastest rates at Kew .
The traces were , however , of a very different type from the others , oscillations of short period being much less in evidence .
Considering that the ranges of the diurnal inequalities in and at Eskdalemuir in December are only about 15 and 20 respectively , it is a little startling to find that during Dr. C. Chree .
six hours of one particular " " day the mean rates of change per minute in these elements were quite .
The contributions to the aggregates oi changes from the other 18 hours of the 24 would have been considerable , and the mean rate of change in and for the whole 24 hours could hardly have been less than 4 per minute .
S6 .
Any complete estimate of the expenditure of energy during a magnetic storm is probably impossible .
A magnetic storm is usually accompanied by aurora and earth currents .
In this particular case , according to observations made by Prof. Stormer near Christiania , the aurora extended from 100 to 400 kiloms .
above the earth 's surface .
There is no known method of finding the intensity at any height of the electric currents producing aurora .
Few stations attempt to measure earth currents , and little , if anything , is known as to the depth to which earth currents extend .
Thus the energy expended in aurora , and the expended in earth currents , are alike inaccessible to calculation .
If the ultimate source of all the terrestrial phenomena is some form of electrical from the sun , the energy represented by the various phenomena in the earth and the atmosphere is probably but an insignificant fraction of the whole .
The estimate of the energy represented by the changes shown in netic curves is naturally a simpler problem , but even it is fraught with difficulties , practical as well as theoretical .
For what has been done to give definiteness to the problem we are mainly indebted to the late Prof. Bidlingmaier .
* The formula given by Maxwell for the energy in a netic field when there are no electric currents is where represent the rectangular components of netic force .
The is supposed to be taken throughout the whole of the magnetic field .
It takes , moreover , as point of departure a total absence of force .
In the present case we know the absolute values of the three components at a fixed point , and the sequence of changes in these components ; but the intensity of the field never vanishes , the changes encountered being in fact small compared with the mean values of the ments .
Suppose we confine our attention for the moment to one of the three components .
The value may be regarded as given by the curve ordinate at the instant .
If be the accepted nonmal value , the ' activity as Bidlingmaier has called it , at the instant is .
( 1 ) Veroffentlichungen des Kaiserlichen Observatoriums in Wilhelmshaven , ' " " Ergebnisse der Magnetischen Beobachtungen im Jahre 1911 Never Folge , Heft 2 .
The Magnetic Storm of December 16-17 , 1917 .
What we are concerned with in practice is the mean value of the " " activity\ldquo ; for some specified time , whether an hour , a day , or a year .
Let , ( 2 ) where is the mean value for the hour under consideration .
Then , from the properties of the arithmetic mean , we have Mean value of per hour mean value of per hour .
( 3 ) To deterluine the mean value of , Bidlingmaier measured the curve at six-minute rals and took 1/ 10 as the mean value for the hour .
In the present case the curve was measured at four-minute , at , , 56 , 60 minutes after the hour , and , ( 4 ) was accepted as the mean .
Thus answers to what Bidlingmaier called for the particular hour .
The mean\ldquo ; activity\ldquo ; for the one hour is Suppose now we want the mean activity for the day .
If be the mean value for the day , and , .
the mean values for the several hours , then Mean value of In Bidlingmaier 's notation , ( 7 ) .
( 8 ) His complete expression for the mean " " activity\ldquo ; of the day is .
( 9 ) In this connection represents the arithmetic mean of the 24 individual hourly values of .
The curve ordinates are to be regarded as expressed in terms of , and the unit in the " " activity\ldquo ; results is 1 erg per cubic centimetre .
A difficulty at once arises as to , viz. , what to accept for the normal value .
Bidlingmaier apparently thought that theoretically the best ] would be to derive it from the mean value for the whole year , by assuming the rate of secular change uniform .
For instance , for Wilhelmshaven in 1911 , taking per annum as the rate of change of , he accepted as the value of for a day days subsequent to July 1 Mean for year His value of thus increased gradually throughout the year .
Dr. C. Chree .
The mean value for a year is unknown until the year is completed .
For the secular it is desirable to know the mean value for the subsequent as well as the previous year .
Thus the calculation of in the way outlined above would have to be postponed to a somewhat indefinite date .
Recognising this , though he gave values of at Wilhelmshaven for all days from January 1 to June 30 , 1911 , the period to which he confined himsel maier concluded that in general the only practical course would be to omit , i.e. , to accept for the mean value of the individual day .
This would give for the day 's mean " " activity\ldquo ; , the symbols having the same meaning as before .
In accepting as the best theoretical value for one based on the mean value for the year , Bidlingmaier recognised one difficulty .
The mean value is not the same when we confine ourselves to quiet days as when we include all days of the year .
This is especially true of H. The quiet day yearly mean would seem the more natural one to employ in the present connection , but that is a point on which opinions may differ .
Another difficulty is that an annual inequality is believed to exist in the case of all the magnetic elements .
If an annual inequality really exists , it should presumably be taken into account in the present problem .
These considerations support the view that the practical difficulties in the way of applying to individual days a correction based on the mean value of the element for the year and the secular change are very serious .
When we examine the results for individual days of 1911 , which Bidlingmaier found at Wilhelmshaven , we find that is by no means trifling .
The following selection from Bidlingmaier 's results should carry conviction on this point:\mdash ; On the average day the value of is here about three-quarters that of , and on some individual days is of the same order as .
Our view as to the relative " " activities\ldquo ; on April 9 and March 20 , two of the most disturbed days of 1911 , for instance , would be largely determined by the retention or neglect of The adoption of Bidlingmaier 's ' activities\ldquo ; as an international scheme , in place of the present ' character\ldquo ; figures , was under consideration before The Magnetic Storm December 16-17 , 1917 .
war .
The question whether a correction of the type of should be applied ought , I think , to receive very careful consideration before any final decision is reached .
If the present method of , the international quiet days continues , perhaps the most satisfactory course would be to accept the mean value from the five quiet days as the for all days of the month .
At present three months are dealt with together , and the announcement of the quiet days selected , even in pre-war times , did not appear until several months had elapsed , after the end of the quarter .
No doubt matters could be accelerated by dealing with the months individually .
If Bidlingmaier 's scheme were substituted for the present scheme , the selection of the quiet days would entail the pre-existence of the " " activity\ldquo ; statistics .
In that event , possibly , the best course might be to accept for the mean value of the last previous day considered quiet at the particular observatory .
S7 .
In the case of a distnrbance so large as that of December 16-17 , 1917 , a difference of or in is of minor importance .
I decided to calculate two sets of " " activity\ldquo ; results .
The first set accepts a mean derived from the " " day\ldquo ; ending at 8 .
on December 16 .
The second set adopts Bidlingmaier 's suggestion , and neglects entirely , accepting for the mean value for the " " day\ldquo ; commencing at 8 .
on December 16 .
For the disturbed day the rqean of the 24 hourly means was taken .
For the previous or quiet day , tho mean was taken from measurements made at 14 .
and 20 .
on the 16th , and at 2 .
and 8 .
on the 17th .
The mean of readings taken at these four hours on the average quiet day agrees very closely with the mean from the 24 hours in all the elements .
With a view to possible future intercomparisons it is desirable to put on record the two sets of mean curve ordinates .
At Eskdalemuir the equivalents of 1 mm. of ordinate were iu in and in .
At Kew the scale value determinations gave Dr. C. Chree .
as the equivalents of 1 mm. of ordinate in in , and 16 .
in In presenting the results obtained from the use of the quiet day mean , I have not followed Bidlingmaier 's method of obtaining a correction This is convenient when we confine ourselves to the mean\ldquo ; activity\ldquo ; for the whole day , but seems less appropriate when we consider individual hours .
In fact it is not altogether clear how Bidlingmaier proposed to treat individual hours , except when is neglected .
In his Table XIII , *when dealing with the diurnal variation of " " activity he takes no cognisance of He gives , however , two sets of figures .
The first set includes mean values obtained each hour of the day from the five international quiet days of the month , the second set gives the excess over these values of the corresponding values obtained when all days of the month are included .
The second set of figures are regarded by Bidlingmaier as giving the diurnal variation of disturbance .
They would obviously be unaffected by any correction which was common to the all-day and quiet-day " " activities I am doubtful whether Bidlingmaier 's method of deducing a diurnal variation of disturbance is satisfactory .
The differences between the curve ordinates at one and the same hour of the five quiet days of a month must represent in the main disturbance , thus it is only what is common to the five days that can well claim to represent entirely quiet conditions .
In the present case , even if I had entirely approved Bidlingmaier 's procedure , it would have been impracticable to adopt it , because the selection of quiet days for December , 1917 , is unlikely to be announced until midsummer , 1918 .
S8 .
The best course seemed to be to refer each hourly measurement directly to the quiet day mean accepted , so that the complete mean value of the " " activity\ldquo ; for the hour becomes .
( 10 ) Tables IX and X give the results thus calculated for individual hours at Kew and Eskdalemuir .
The unit employed in the Tables is that used by maier , viz. , erg per cubic centimetre .
Bidlingmaier 's notation being complicated , the following notation has been employed in the Tables and their discussion .
( 11 ) .
cit. , p. 29 .
The Magnetic Storm of December ] Here has the same meaning as in ( 2 ) , while are the mean values for the days which end and commence respectively at 8 .
on December 16 .
The excess in the mean value of for the day over that of is equivalent to Bidlingmaier 's .
The values derived for the from ( 8 ) , acceptin the differences between .
and given in S7 , are as follows , in terms of the unit employed in the Tables : At Kew Observatory in in in At Eskdalemuir Observatory in in , 303 in All the preliminary arithmetical operations which went to the construction of the Tables were performed on differences of curve ordinates expressed in millimetres or minutes of arc , the ulit 1 erg being introduced only in the final operation ; thus cm absolute identity was not to be expected between the differences between and just calculated , and those derived in Tables IX and X by taking the actual means of the 24 hourly values .
It mtty be well to add that a calculation was made to ascertain the order of magnitude of the corrections that would have been applied if an attempt had been made to allow for the " " activity\ldquo ; present on quiet days .
The calculations were applied to the inequalities for December as given by five selected quiet days a month .
The Kew data were for the 11 years 1890-1900 combined , the Eskdalemuir data for the single year 1914 .
The means of the 24 hourly values thus found for ( Bidlingmaier 's ) were in the usual unit : At Kew , for ; for ; for ; total for 3 elements ; At Eskdalemuir , for ; for , for ; total for 3 elements No data were available for the calculation of A2 ( or ) , but the contributions from on a quiet December day must be very small .
If the quiet days had been treated individually , larger mean values would have been found for .
But , if we may judge from Bidlingmaier 's data for January , the whole corrections with the quiet days treated individually would not have exceeded double the figures given above , and the largest correction for any individual hour would have been only from three to four times the mean .
The effect of any such corrections in Tables IX and X would have been The calculations were carried to the first decimal place ; but it has been omitted in all the values of the total " " activities as well as in all individual cases where the value of , or exceeded 100 .
Results are given for the two horizontal components combined , as well as for the sum from the .
cit. , p. 29 .
VOL. XCIV. .
2 Dr. C. Ch.ree .
The gnetic Storm of December , 1917 .
Dr. C. Chree .
three components , because it was the only way of intercomparing the Kew and .
results from the horizontal plane , and because Bidlingmaier omitted vertical force .
The relative importance of and is of special interest in connection with the question of the calculation of , a very laborious process when carried out strictly according to the formula .
If we express A2 as a percentage of in the case of the mean values for the day , we obtain the following results:\mdash ; At Kew , in , 11 ; in , 9 ; in , 4 ; in , 10 ; in , 8 ; At Eskdalemuir , in , 21 ; in , 11 ; in , 6 ; in , 15 ; in .
With these we may coml ) Bidlingmaier 's mean results at Wilhelmshaven for January to June , 1911 ; , in , 24 ; in , 21 ; in , 22 .
It would appear , however , that most of Bidlingmaier 's results were derived not from actual measurement of , but from measurements of the hourly ranges , and certain numerical relationships which he believed to hold between the value of and the square of the corresponding hourly range .
In a recent paper , arrived at the conclusion that Bidlingmaier 's relationships were not generally satisfactory .
If they had been accepted in the present case , we should unquestionably have got a higher mean value for .
If were always as small as the above values for Kew suggest , it could not be claimed that any very high degree of accuracy is essential in the method of calculating Bidlingmaier found that on the average day the\ldquo ; activity whether or , was considerably greater for than for H. On the present occasion , the " " activity\ldquo ; at Kew , especially and , is conspicuously larger for than for D. The and results for and at Eskdalemuir differ less , which suggests that the horizontal direction in which the changes of force were most active made a fairly close approach to the netic meridian .
In temperate latitudes the range of the regular diurnal inequality in is considerably less than in either horizontal component .
] , undoubtedly , on the average day at any European station , the contribution from will be small compared with the sums of the contributions from the horizontal components .
But , on the present occasion , the contribution from to or at Kew is nearly equal to the combined contributions from and , while at Eskdalemuir the contribution from considerably exceeds that from and combined .
This shows that the question of the 'Terrestrial Magnetism and Atmospheric Electricity , ' , p. 67 ( 1917 ) . .
The Magnetic Storm of December 16-17 , 1917 .
545 exclusion of data from an international scheme possesses an importance which is not suggested by a study confined to ordinary days .
S9 .
It is interesting to note , confining ourselves to the results for that the mean value of at Kew on December 16-17 was times that of the average day from January 1 to June 30 , 1911 , at Wilhelmshaven , and 15 times that of the average January day .
Also , the mean value of at Kew between 21 .
and 22 .
on December 16 was 52 times that of the corresponding hour of the average January day at Wilhelmshaven . !
It is obvious that , when we confine ourselves to data from a single year , few highly rbed days exercise a swamping influence in statistics as to the annual or diurnal variation of magnetic " " activity During the earlier part of the storm the " " aCtiyity\ldquo ; at was not inferior to that at Eskdalemuir , but after 16 .
on the 16th the Eskdalemuir " " activity\ldquo ; was much the greater .
At both stations , so far as the horizontal field is concerned , 21 h.-22 .
on the 16th was decidedly the most disturbed hour .
This was the time at which aurora was ightest , according to reports from meteorological stations , several Scottish stations reporting a corona not far from the zenith .
Kew and Eskdalemuir also agree in showing a maximum of " " activity\ldquo ; in between 17 .
and 20 .
on the 16th , and a marked recrudescence of " " activity\ldquo ; between 2 .
and 4 .
on the 17th , which was presumably associated with auroral streamers at Southport between .
and 2S S10 .
In view of the great labour involved in applying Bidlingmaier 's ' scheme , and the various uncertainties , I yested in ' Terrestrial Magnetism ' ( loc. cit. ) that a much simpler scheme , if applied at the numerous observatories which participate in the present international scheme , suffice for the magnetic classification of days .
The suggestion was that , if represent the absolute ranges in the three netic elements , or , if the horizontal field only is considered , might be accepted as a measure of the mean " " activity\ldquo ; for the day .
If the horizontal components recorded were and , then would take the place of I have compared the mean monthly values of the " " activity\ldquo ; at Wilhelmshaven for three of the six months dealt with by Bidlingmaier , with the corresponding arithmetic means of the squares of the Kew daily ranges .
The following were the results obtained:\mdash ; The Magnetic Storm of December 16-17 , 1917 .
now the " " activities\ldquo ; and ranges found for December 16-17 , we get the following results:\mdash ; At Kew\mdash ; ( Mean for ) Mean for for ) Mean for At ( Mean for ) Mean for ( Mean for ) Mean for The difference between the above figures for the two stations arises from and .
If we confined ourselves to we should get in place of the above ratios : For the horizontal components , at ICew , at Eskdalemuir ; for all three components , at both places .
The means of the absolute daily ranges for the whole year 1914 were At Kew , in , and in ; At Eskdalemuir , in , and in W. This makes Wilhelmshaven being very nearly midway as regards latitude between Kew and Eskdalemuir , the absolute ranges at Wilhelmshaven probably exceed those at Kew .
Thus if we had employed Wilhelmshaven instead of Kew ranges for comparison with the Wilhelmshaven ' activities\ldquo ; of 1911 , we should most likely have got smaller values than those obtained above for the ratios .
The scheme which I suggested does not postulate that the ratio borne by aotivities\ldquo ; to squares of ranges should be the same for all stations .
It might well be different at a quiet station like Helwan , and at a disturbed A Jfethod of at station like Sitka .
Everything considered , however , it is probable that on any scheme which assumed " " activity\ldquo ; proportional to the square of the absolute range , the mean " " activity\ldquo ; for December 16-17 would have been sensibly over-estimated at Kew , and considerably over-estimated at !
Eskdalemuir .
On the other hand , the difference between the values of and in Table IX , and the uncertainty this implies , shows even with the elaborate procedure entailed by Bidlingmaier 's full scheme , there is a large probable error at stations .
Under these circumstances a simple scheme , even if lmittedly imperfect , desel'ves consideration .
A Method of Avoiding Collision Sea .
By J. JoLY , Sc. D. , F.B.S. , a Commissioner of Iris Light , PART I. ( Received May 8 , 1918 .
) The following method of avoiding collision at sea depends on the use of synchronised signals , transmitted in different media .
Such signals , at different rates , enable the distance of their source to be inferred by observation of the gain in time of the faster upon the slower travelling signal .
Thus , if signals be simultaneously emitted by wireless and by submarine bell ( or Fessenden oscillator ) , the former being transmitted with practically infinite velocity , the latter arrive with a which 1s the time the submarine sound requires to traverse the intervening medium .
The rate of propagation of sound iu water closely 4800 feet per second , the lag is second for one-half sea-mile .
In practice the nals may be so ordered as to dispense with the stop- watch or chronograph .
This is accomplished by sending out the wireless ticks in groups of , say , 20 " " dots\ldquo ; spaced to intervals of second .
The stroke of the bell precedes the first of these dots by one of these intervals .
Thus , when the sailor is half mile frdm the source he hears the first wireless dot along with the bell stroke .
If he is 1 mile distant the bell stroke comes in with the second dot , and so on .
He has , in fact , only to count up the dots till he hears the bell , and the number of the dot coincident with the bell is the number of half sea-miles intervening between his ship and the source of the signals .
It is possible to estimate the quarter mile by noting a want

