43 Barometric Variations of Long Duration over Large Areas .
By William J. S. Lockyer , M.A. , Ph. D. , F.R.A.S. , Chief Assistant , Solar Physics Observatory , South Kensington .
( Communicated by Sir Norman Lockyer , K.C.B. , LL. D. , Sc. D. , F.R.S. Received May 8 , \#151 ; Read June 21 , 1906 .
) [ Plates 1\#151 ; 5 .
] In a paper* communicated to the Society in the year 1902 , Sir Norman Lockyer and I pointed out the existence of a barometric see-saw of short durationf ( about 3'8 years ) occurring between two large regions , nearly antipodal to each other , the centres of which were approximately India and Cordoba ( South America ) .
A continuation of the research indicated that this barometric see-saw was of greater extent than was at first supposed , and in a further communication !
in the same year observations extending over new regions were discussed and the results published .
Still further inquiry indicated that this see-saw was almost world-wide in its extent , and the result of a later investigation , which included the examination of pressure observations at 95 stations scattered over the earth 's surface , was communicatedS to the Society in the year 1904 .
During the progress of this work it was noticed that many of the curves representing barometric changes over a great number of years in various widely distributed areas exhibited variations of much longer duration which were neither coincident in their epochs of maxima or minima or even in the lengths of their oscillations .
It seemed to me therefore quite possible that if a distinct pressure change of short duration could occur simultaneously , but of opposite phase , in anti* ' Roy .
Soc. Proc. , ' vol. 70 , p. 500 .
+ The use of the word period previously employed in these investigations did not necessarily imply that the curves discussed were perfectly " " but that the meteorological variations which they represented showed oscillations which , when more will be known about them , may possibly be found to be dependent on a periodic cause or causes .
The sun-spot variation , for instance , is generally described as being periodic , with a period of 11T years .
As a matter of fact this is only a mean value of the intervals , minimum to minimum , the departures from this mean amounting sometimes to two years .
Such differences from the mean may be due to other underlying periods , one of which of about 35 years in length was suggested in my paper entitled " The Solar Activity 1833\#151 ; 1900 , " communicated to the Royal Society in 1901 , ' Roy .
Soc. Proc. , ' vol. 68 , p. 294 .
| 'Roy .
Soc. Proc. , ' vol. 71 , p. 134 .
S ' Roy .
Soc. Proc. , ' vol. 73 , p. 457 .
Dr. W. J. S. Lockyer .
Barometric [ May 8 , podal parts of the earth it was not unreasonable to expect variations of long duration to behave in a like manner .
The question was , however , well worth inquiring into , and the object of this communication is to present to the Society the first results of the investigation .
In the present instance the survey does not extend over the whole world , but is restricted to the areas which include India , the East Indies , Australia and South America .
It may be stated further that , for such an investigation as this , as long a series of homogeneous observations as possible for each station is desirable .
The available data are , however , not so numerous as would have been wished , but in spite of this I think that the results deduced are sufficiently consistent to serve as a first approximation to the barometric changes in operation in these areas .
The paucity of barometric data is most conspicuous in the South American region , so that for this part of the world the change of long duration has to be deduced from several portions of series of observations made at various stations .
This I am aware is not a very satisfactory procedure , but failing this no other means is available .
In a previous communication* curves have been published representing barometric changes of short duration for Bombay , Cordoba and Adelaide , commencing in the years 1873 , 1873 and 1876 respectively .
Since the curves for Bombay and Adelaide can be carried much further back than the above dates , namely to 1847 and 1857 respectively , I have considered it necessary and advantageous to bring together in the first instance a series of curves 'representing the short period variation for the regions dealt with in this paper .
Further , two or more independent curves are given for each area investigated , in order to demonstrate the real nature of such variations .
Plate 1 gives seven curves exemplifying the variations of the mean annual pressure changes from year to year .
Madras and Bombay represent India , Batavia the East Indies , Melbourne and Adelaide are typical of Australia , and Cordoba and Santiago show the changes in the South American region .
Since the barometric variation of short duration of the last two mentioned places is of opposite phase to that occurring in the Indian area , these two curves representing their changes have been inverted in the plate .
The curves are all drawn on the same scale , so that a quantitative as well as a qualitative idea may be obtained of the amplitudes of the variations in these different and widely distributed regions .
It is only necessary to take a rapid glance at this series of curves to form an idea of the very close relationship that exists between the barometric * 'Roy .
Soc. Proc. , ' vol. 73 , Plate 13 .
1906 .
] Variations of Long Duration over Large Areas .
changes in these parts of the world .
As the curves indicate , the changes are of the greatest magnitude .
in the Australian area , while in the East Indies , India and South America , they are more or less of the same order of intensity .
This magnitude , although for a great part dependent on the geographical positions of the areas as regards latitude , is not apparently entirely so , since there is a great difference between the values for the Cordoba and South Australian stations , the latitudes of which are approximately the same .
In a subsequent paragraph reference will be made to the effect of latitude on the percentage values of the amplitudes of the variation of short duration in terms of the mean annual swing .
To determine approximate relative magnitudes of the largest amplitudes of these changes of short duration for these regions , the five most prominent rises in the curves from a minimum to a succeeding maximum have been selected and the differences of the readings determined .
In the case of Madras and Bombay the means are respectively 0*033 and 0*036 inch , which give for the Indian area a mean maximum amplitude of 0*035 inch .
The individual years taken ancf1 the corresponding barometric readings were as follows Station .
Year .
Minimum .
Year .
Succeeding | maximum .
Difference .
inches .
inches .
Bombay 1862 29 -778 1864 29 -827 1875 29 -812 1877 29 -848 1878 29 -801 1881 29 -825 1886 29 -812 1888 29 -836 .
1898 29 -795 1899 29 -830 Mean ... 29 -800 ... 29 -833 0*033 Madras 1862 29 -806 1864 29 -851 1875 29 -814 1877 29 -866 1882 29 -806 1884 29 -840 1887 29 -810 1888 29 -836 1898 29 -805 1899 29 -828 Mean r - 29-808 | 29*844 0*036 For three stations in Australia the values for the amplitudes , derived in a similar manner , are 0*076 inch for Melbourne , 0*077 inch for Adelaide and 0*071 inch for Sydney , the mean value for Australia being 0*074 inch .
The values from which these means have been derived are as follows ( see p. 46 ) .
Comparing the means for the Indian and Australian regions , it is seen that the ratio of amplitudes is as 35 to 74 , that is as 1 to 2 nearly .
Dr. W. J. S. Lockyer .
Barometric [ May 8 , Station .
Year .
Minimum .
Year .
Succeeding maximum .
Difference .
inches .
inches .
Melbourne 1863 29 -896 1866 29 -954 1875 29 -886 1877 29 -993 1878 29 -905 1881 29 -966 1882 29 -902 1885 29 -996 1890 29 -924 1891 29 -985 Mean ... 29 -903 ... 29 -979 0-076 Adelaide 1863 30 '006 1865 30 -073 1875 30 -028 1877 30 -144 1879 30 -052 1881 30 -107 1882 30 -047 1885 30 -121 1890 30 -036 1891 30 -111 Mean ... 30 -034 ... 30 -111 0-077 Sydney 1867 29 -872 1868 29 -933 1875 29 -803 1877 29 -896 1879 29 -816 1881 29 -874 1882 29 -829 1885 29-920 1893 29 .
829 * 1894 29 -885 Mean ... 29 -830 ... 29-901 0-071 General mean ... \#151 ; ... 1 ... 0-074 This fact seems to me of very great importance , because , if these barometric changes of short duration are on so much larger a scale in Australia than they are in India , a study of the Australian conditions may materially assist Indian Meteorologists .
The amplitude for the South American region is of about the same order as that of India , the mean for five stations derived in the above-mentioned manner being 0'38 inch .
The individual stations and corresponding values are as follows- ( see opposite page ) .
The first step taken to render more apparent the changes of long duration involved in all these curves was to eliminate as far as possible the prominent short variations of about four years ' duration .
This was to a great extent accomplished by grouping the years in sets of four and employing the mean values of each of these groups .
Thus the means for the years 1873 to 1876 1874 to 1877 , 1875 to 1878 , and so on , were determined , and curves were drawn through each of these points after they had been plotted on squared paper .
Each mean point was actually plotted on the time scale at the end of the second year of the group of which it was the mean .
Thus the mean for 1873 to 1876 was plotted at the end of 1874 .
The curves first dealt with in this manner were those showing the pressure changes for India .
Although one curve would have been sufficient to 1906 .
] Variations of Long Duration over Large Areas .
illustrate the variation over this area , a second curve is added to serve as a check on the accuracy of the first .
For this region the barometric observations made at Bombay and Madras were utilised since they go back so far as 1847 and 1842 respectively .
Station .
Year .
Minimum .
Year .
Succeeding maximum .
Difference .
inches .
inches .
Cordoba 1873 28 -494 1875 28 -534 1884 28 -498 1886 28 -532 1888 28 -499 1890 28 -531 1891 28 -508 1893 28 -552 1902 28 -487 1903 28 -522 Mean ... 28 -497 ... 28 *534 0-037 Curityba 1887 27-008 1888 27 -043 1891 27 -028 1894 27 055 1899 27 -031 1900 27 -055 1902 27 -031 1903 27 -057 Mean ... 27 -025 ... 27 -053 0-028 Goya 1877 29 -782 1878 29 -832 1888 29 -805 1890 29 -840 1891 29 -828 1893 29 -858 Mean ... 29 *805 ... 29 -843 0-038 San Juan ( Buenos 1868 29 -892 1871 29 943 Airs ) 1873 29 -930 1874 29 -979 1877 29 -906 1878 29 -946 1887 29-880 1889 29 -923 1891 29 -908 1893 29 -954 Mean ... 29 -903 ... 29 -949 0-046 Santiago 1871 28 -235 1872 28 -269 1877 28 -185 1879 28 -235 1888 28 -217 1890 28 -238 1900 28 -176 1901 28 -232 1902 28 -182 1903 28 -230 Mean ... 28 -199 ... 28 -241 0-042 General mean ... ... ... ... ... 0-038 These two curves are reproduced on Plate 2 and it will be seen that they are very nearly identical in every respect .
They exhibit , further , variations of a periodic nature , but this periodicity becomes muoh less apparent after the year 1880 .
The next curve examined was that of Batavia ( East Indies ) .
Unfortunately the observations do not go back further than 1866 , but where the curve overlaps those of India the close similarity of the changes is very apparent .
4.8 Dr. W. J. S. Lockyer .
Barometric [ May 8 , Going still further away from India , the Australian continent was then examined .
Here the curves representing the variation of long duration begin to present a new aspect .
Fortunately three excellent series of barometric observations are available for this region , namely , those for Adelaide , Melbourne , and Sydney , commencing respectively in the years 1857 , 1859 , and 1858 .
The changes atj Perth are also included here , as there is a set of observations commencing ' iji 1875 which represent the variations taking place well to the west of this continent .
... ... .
All these four curves are reproduced on the scale in Plate 2 .
It will be seen in the first instance that the Adelaide curve resembles in a general way that of Bombay and that the maximum about the years 1877 and 1878 is almost equally pronounced in Adelaide .
Attention is specially drawn to this particular maximum , as it will be observed when examining the curves for Melbourne , Sydney , and Perth , that during these years it becomes of quite secondary importance .
If for the moment the curves at this epoch be left out of consideration , it will be seen that the remaining portions of all the curves are not only very like each other , but bear a general resemblance to those of the Indian region .
Before dealing with the possible origins of these barometric variations of long duration in the eastern hemisphere , attention will first be directed to ai part of the western hemisphere , namely , that of South America , to see whether similar changes are in operation there , and if so , to study their \#166 ; nature .
It ; will be remembered that the barometric change of short duration in this region behaved in an inverse manner to that of the Indian and Australian areas ( see Plate 1 ) .
Unfortunately for this part of the world data are not numerous , but still , I think they are sufficient on the present occasion not only to demonstrate that a long barometric variation does take place , but that the epochs of maxima and minima are not those of either India or Australia as deduced in this paper .
Employing the same method as above described , four-year means were determined and curves drawn for the five stations , Cordoba , Goya , and San Juan ( Argentine Republic ) , Santiago ( Chili ) , and Curityba ( Brazil ) .
All these curves are reproduced in Plate 8 , and are drawn on the same scale as those of the Indian and Australian curves in Plate 2 .
Although the South American curves extend over different periods of time , there is sufficient overlapping in all cases to connect up one series with another . . .
_ : L906 .
] Variations of Long Duration over Large Areas .
49 The Cordoba curve undoubtedly indicates that a long barometric change is taking place , but the shortness of the period over which the observations extend , namely , from 1873 to 1904 , renders it unserviceable for the determination of its possible periodicity .
A neighbouring station , Goya , corroborates in a general manner , so far as the observations extend , the Cordoba variation , with perhaps the exception of the first three points on the curve .
To carry back these pressure changes to an earlier date , the observations at San Juan ( Buenos Airs ) were employed ; the available data for this station extend from 1867 to 1889 .
Here we find the fall of pressure at Cordoba from 1875 to 1882 well corroborated , followed by a subsidiary maximum similar to that at Goya in 1885 .
So far as these observations extend , there seem to be two prominent maxima at about the epochs 1874 and 1893 , each of which is followed by minima at about the years 1882 and 1901 .
The curve for Santiago , a station to the west of the Andes , indicates also very clearly these two principal maxima and the second of the two minima at the same epochs , but the minimum about the year 1882 occurs somewhat earlier .
At a station in Brazil , Curityba , in which only a short series of observations is available , this long variation is also in existence ; the second principal maximum , however , falls a little later than at the previously mentioned stations .
In order to indicate generally the approximate dates of the occurrence of the points of maxima and minima in these four-year mean curves , the following table has been drawn up .
In this the large type figures represent the years of the most prominent maxima and minima , while those in smaller type indicate those of a subsidiary or uncertain nature:\#151 ; Years of maxima .
Years of minima .
India .
Batavia .
Adelaide .
Perth , Melbourne , Sydney .
South America .
India .
Batavia .
Adelaide .
Perth , Melbourne , Sydney .
South America .
1845 .
1857 - _____ 1849 \#151 ; \#151 ; \#151 ; \#151 ; 1867 1867 1868 1862 \#151 ; 1862 1863 \#151 ; 1877 1877 1878 1878 1874 1871 1872 1871 1875 \#151 ; 1886 ?
1884 ?
1887 1887 1881 1881 1881 1881 1882 1901 ?
1901 ?
( 1901 ?
) \#151 ; 1893 1894 1894 1894 1894 1901 VOL. lxxviii.\#151 ; A , E 50 Dr. W. J. S. Lockyer .
Barometric [ May 8 , Generally speaking , the above figures indicate that while the Indian and Batavian ( and Adelaide ) curves show a variation , the duration of which is somewhere about 11 years , the curves for the other Australian stations and those of South America present a change in which the maxima and minima are about 19 years apart .
Reference to these changes will be made more fully in subsequent paragraphs .
It is important further to note , as was done in the case of the variation of short duration , the very great difference between the amplitudes of these long changes in the Indian , Australian , and South American regions .
If for instance the difference between the readings of the several maxima and minima of the curves plotted for Bombay and Madras be determined , it is found that they are 0*026 and 0'026 inch respectively , the mean of which is 0*026 inch .
For Adelaide , Melbourne , and Sydney the differences for each are 0*054 , 0*043 , and 0*058 inch respectively , the mean of these three being 0*052 inch .
The actual values from which the above figures have been derived are given in the following tables , but since in this instance the points on the Bombay .
Madras .
Year .
Minima .
Mean .
1848\#151 ; 49 1861\#151 ; 62 1870\#151 ; 71 inches .
29 -808 29 *808 29 *804 29 *807 Maxima .
1856\#151 ; 57 1865\#151 ; 66 1876\#151 ; 77 29 *832 29 *837 29 *830 29 *833 Difference ... ... 0*026 Tear .
Minima .
Mean .
1861\#151 ; 62 1871\#151 ; 72 1893-94 inches .
29 *789 29 *797 29 *805 29*797 Maxima .
1875\#151 ; 76 1886-87 1900\#151 ; 01 29 *821 29 *824 29 *823 29 *823 Difference ... ... 1 0 *026 1 General mean difference 0 *026 inch .
Adelaide .
Melbourne .
Year .
Minima .
Mean .
1861\#151 ; 62 1870\#151 ; 71 inches .
30 -023 30 *035 30 *029 Maxima .
1877\#151 ; 78 1886\#151 ; 87 1895\#151 ; 96 30-082 30 *093 30 -073 30 *083 Difference ... ... 0-054 Year .
Minima .
Mean .
inches .
1862\#151 ; 63 29 -914 1873\#151 ; 74 29 *921 1893\#151 ; 94 29 -915 29 *917 Maxima .
1867\#151 ; 68 29 -947 1886\#151 ; 87 29 *973 29 *960 Difference ... ... 0*043 1906 .
] Variations of Long Duration over Large Areas .
51 Sydney .
Year .
Minima .
Mean .
1862\#151 ; 63 1874\#151 ; 75 1880\#151 ; 81 inches .
29 -842 29 -829 29 -837 29 -836 Maxima .
1857-58 1886\#151 ; 87 1895\#151 ; 96 29 -896 29 -902 29 -883 29 -894 Difference ... ... 0-058 General mean difference 0*052 inch .
curve are derived from means of four years , the years given refer to the two middle years of each four .
In the case of Australia then this long variation has an amplitude twice as large as that of India , a feature which I think has not been pointed out previously .
The magnitude of the amplitude of this long barometric swing must have a great effect in changes in Australian weather , and most probably is responsible for the marked secular weather changes which have been previously shown to exist in this region of the world .
To determine the amplitude of the oscillation in South America , differences between the readings of the maxima and minima points on the four-year mean curves were also determined .
The mean of the values for Cordoba ( 0-032 inch ) , Santiago ( 0-045 inch ) , and San Juan ( Buenos Airs ) ( 0-039 inch ) was 0'039 inch .
This value is intermediate between the Indian ( 0"026 inch ) and the Australian ( 0'052 inch ) amplitudes .
The values for the individual stations were as follows:\#151 ; Cordoba .
Santiago .
Year .
Minima .
Mean .
1881\#151 ; 82 1900\#151 ; 01 inches .
28 -499 28 -492 28-495 1875\#151 ; 76 1893\#151 ; 94 Maxima .
28 -521 28 -524 28 -527 Difference ... ... 0-032 Year .
Minima .
Mean .
1876\#151 ; 77 1900\#151 ; 01 ' inches .
28 -215 28 -192 28 -203 Maxima .
1872\#151 ; 73 1893\#151 ; 94 28 -252 28 -244 28 -248 Difference ... ... 0-045 e 2 Dr. W. J. S. Lockyer .
Barometric San Juan ( Buenos Airs ) .
[ May 8 , Year .
Minima .
Mean .
1881\#151 ; 82 1887\#151 ; 88 inches .
29 -911 29 909 29 -910 Maxima .
1873\#151 ; 74 29 -949 29 -949 Difference ... 0-039 General mean difference 0 *039 inch .
It may be of interest here to summarise in tabular form the amplitudes of the mean annual and the short and long variations of barometric pressure for the stations used in this enquiry which have been found to be most conspicuous .
Two columns of figures , the third and last , indicate the percentages of the amplitudes of the short and long changes in terms of the mean annual variations determined from a long series of years in each case:\#151 ; Amplitudes of Pressure Variations .
Stations .
Mean annual variation .
Short variation .
Percentage of annual .
Long variation .
Percentage of annual .
Bombay 0-283 0-033 11 -7 0-026 9-2 Madras 0-297 0*036 12 -1 0-026 8-8 Adelaide 0-226 0-077 34-0 0-054 23 1 Melbourne 0-204 0-076 37-2 0 043 21 -1 Sydney 0-212 0-071 33 -0 0-058 27-3 Cordoba ... .* 0-177 0-037 20 -9 0-032 18 -1 Curityba 0-188 0 -028 14 -9 \#151 ; \#151 ; Goya 0-248 0-038 15 -3 San Juan ( Buenos Ayres ) 0-236 0-046 19 -5 0-039 16 -5 Santiago 0*128 0 042 32 -7 0-045 35 -1 The values shown in this table are very striking .
In the case of Australia , for example , the fact is made apparent that the amplitude of the variation of short duration amounts , in the mean to as much as 35 per cent , of that of the mean annual variation , while that of the variation extending over about 19 years reaches nearly 25 per cent. Considering the great magnitudes of these changes in relation to those of the mean annual variations , no doubt can remain as to the important rdle which must be played by them in bringing about changes in the seasons from year to year .
1906 .
] Variations of Long Duration over Large Areas .
53 In a previous paragraph it has been mentioned that the amplitudes of the variation of short duration , expressed in percentage of the annual swing , suffers a change on account of the position of the stations as regards latitude .
To illustrate this for the region including India and Ceylon , the pressure observations at several stations have been analysed , and the result is given in the following table .
The stations are arranged in order of latitude , so that the increase in latitude will be seen to correspond generally with a decrease in the values of the percentages mentioned above:\#151 ; \#151 ; Lat. N. Mean annual variation .
Short variation .
Percentage .
Colombo o / 6 56 inch .
0-094 inch .
0-027 28-7 Coimbatore 11 0 0-200 0 036 18 -0 Madras 13 4 0-297 0-036 12 1 Bombay 18 54 0-283 0-033 11 -7 Nagpur 21 9 0-436 0-036 8-3 Allahabad 25 26 0*555 0-031 5 -6 Jaipur 26 55 0-498 0-025 5-0 Jacobabad 28 24 0-647 0-031 4-8 Lahore 31 34 0 -574 0 039 6-7 D. Ismail Khan 32 0 0-620 o-oio 6-4 This percentage change is not so much due to the variation of the amplitude of the change of short duration as to the increase with the latitude of that of the mean annual variation as shown in the second column of figures above .
The question now arises : how does the long South American variation compare with those shown to exist in the Indian and Australian areas ?
In the first instance it will at once be seen that the South American curve is of a type conforming more , in length of swing , to the Australian variation than to that in India .
Further , there is no doubt about the non-coincidence of the principal epochs .
In order to make a closer comparative study of these variations , hypothetical curves embodying the main features of these changes have been drawn at the bottom of each of these two series of curves illustrated in Plates 2 and 3 .
These are intended to indicate in curves the general nature of the variations as regards their epochs of maxima and minima .
In Plate 2 the epochs of the two principal maxima for the Australian area are seen to occur at about 1868 and 1887 , while three subsidiary maxima are suggested at about the years ( 1858 ?
) 1878 and ( 1898 ?
) , but the epochs of the first and last of these are very uncertain , as the curves do not extend over a sufficiently long period of time .
Nevertheless , it may here be 54 Dr. W. J. S. Lockyer .
Barometric [ May 8 , remarked that the interval between the two chief maxima is 19 years , while those between the successive secondary maxima are about the same length .
Forming a hypothetical curve in exactly the same way for the South American region ( see Plate 3 ) , the two principal maxima here fall in the years 1874 and 1893 , while a subsidiary maximum occurs somewhere between 1880 and 1885 .
Here again the interval between the two main maxima is 19 years .
In both of the hypothetical curves those portions representing the fall from and the rise to a principal maximum have been connected by a dotted line as if a subsidiary maximum did not exist , thus forming a principal ( but really non-existent ) minimum .
The object of doing this is to indicate that in the Australian area the rise to the principal maxima seems to be more abrupt than the fall from them , while in the South American area the opposite feature seems to be the case .
An unsymmetrical curve seemed in both cases to represent the main features better than one drawn symmetrically .
In fact , in the Australian area there is suggested an 8-year rise and an 11-year fall , while in the South American region an 11-year rise and an 8-year fall is indicated .
Particular attention is called to this unsymmetrical peculiarity of the curves , since a similar feature was found to be present in the curve representing the barometric variation of about four years ' duration , * in operation in India and Cordoba .
It was there stated that for Cordoba " the points of maxima of the hypothetical curve at the top of the plate do not lie midway between the minima on either side of them , but nearer the preceding minimum .
" In order to be able to make a direct comparison between these two hypothetical curves representing the barometric changes of long duration in the Australian and South American areas Plate 4 has been added .
In this the South American curve is placed directly above that for Australia , and below the latter is given the South American curve inverted .
The first striking fact which this comparison indicates is the remarkable similarity of the nature of the variation in the two cases .
Both curves seem to have principal maxima occurring at intervals of about 19 years , while situated between these is another maximum of a subsidiary nature .
The second point of importance is that the epochs of these maxima in these two areas are not coincident .
Further , we are not here in the presence of a barometric see-saw , or opposite pressure variation , because the Australian maxima do not occur simultaneously with the South American minima ; there seems to be a general time difference of phase amounting to about * 'Boy .
Soc. Proc. , ' vol. 76 , A , p. 503 , 1905.\#151 ; Note .
1906 .
] Variations of Long Duration over Large Areas .
six years , the epochs of the Australian high pressures preceding those of the South American region .
In the case of the barometric variations of short duration existing between India and South America , the inversion of the latter curve corresponded exactly with the direct curve of the former .
In order to make a similar comparison , the South American curve representing the curve of long duration has here also been inverted and it will be observed that the curves are not the inverse of each other .
The above discussion of barometric changes occurring over large areas shows , I think , clearly that the type of variation taking place in South America is closely similar to that existent in Australia , but unlike that in operation in the Indian area .
With so short a length of time covered by the observations it is difficult to say , especially with regard to Australia , whether the recurrence of the maxima and minima is of a regular periodic nature or not .
Judging by the curves and figures which have been referred to in this paper , there seems to be a general indication that the intervals between the maxima or minima in the case of India are , on the average , approximately 11 years in length , while in the case of the Australian and South American regions the principal maxima are 19 years apart .
I have previously indicated that the Australian pressure change of long duration is undoubtedly very closely associated with that in India , but that the reduction of the intensity of the maximum about the year 1877 in most of the Australian stations alters very considerably the aspect of the Australian curves .
It seems , therefore , probable that if a satisfactory cause be found which is producing the fluctuations in one of these areas , then those in the other area will be subject to the same influence , but will be modified by some cause due possibly to a different geographical position .
In comparing such regions as India and Australia it is as well , in this connection , to bear in mind the continental condition of the former and the insular condition of the latter , each of which has its own special effects on meteorological changes .
In looking for the cause of these barometric changes which extend over several years , it seems as if the solar changes , as exhibited by the frequency or areas of sun-spots ( the only indication of solar activity extending over a long period of time that exists ) , are responsible for the Indian fluctuations .
To indicate this relationship the sun-spot curve ( inverted ) is placed at the top of Plate 2 .
This curve represents the variation from year to year of the mean daily areas of sun-spots deduced from both hemispheres of the sun .
56 Dr. W. J. S. Lockyer .
Barometric [ May 8 , Perhaps different solar data handled in another manner might indicate a closer relationship than is at present suggested .
Although the existing relationship may be considered of too approximate a character to indicate clearly a cause and effect , there is undoubtedly a general similarity between the sun-spot variation curve and that representing barometric changes in India from 1844 to 1903 , a period of 59 years .
Years of average high pressure are years of few sun-spots , and vice versd , but there is a marked exception to this about the epoch of sun-spot maximum in 1883 , which maximum , it may be remarked , was much smaller in intensity than those of 1870 and 1860 .
If India be thus dominated by the solar changes , then the curves for Australia and South America become of secondary importance from the solar point of view and may be considered as a modification of the Indian variation , due possibly to some terrestrial cause .
How this modification is brought about I am not yet prepared to say , but I do not think we need be driven to explain the Australian or the South American barometric changes as depending either on lunar influence or a solar variation of about 19 years .
The similarity of the curves representing the pressure changes in India and the sun-spot curve is not pointed out here for the first time .
In fact , so striking was the resemblance between curves representing these changes in years previous to 1880 that the attention of several meteorologists was drawn to the close association of these two phenomena .
Thus F. Chambers , writing in 1878 , * concluded that the curves " support each other in showing a low pressure about the time of sun-spot maximum and a high pressure at the time of sun-spot minimum .
" He further stated:\#151 ; " The range of the variation of the year by mean pressure from the minimum of 1862 to the maximum of 1868 is 0042 inch and the mean range of the barometer from January to July is 0291 inch , from which it appears that the variation of pressure produced by the absolute variations of the sun 's heat are , in comparison with the usual seasonal changes , by no means insignificant .
" J. A. Brown , f S. A. Hill , J Sir John Eliot , S H. F. Blanford , || E. Douglas Archibalds and others have all corroborated in a general manner this relationship between pressure change and sun-spot variation ; Douglas * ' Nature , ' vol. 18 , p. 568 .
t 'Nature , ' vol. 19 , p. 7 , 1878 .
f 'Nature , ' vol. 19 , p. 432 , 1878 .
S 'Indian Meteorological Keports , ' p. 170 , 1877 .
|| 'Nature , ' vol. 21 , p. 479 , 1880 .
IT ' Indian Meteorological Memoirs , ' vol. 9 , p. 543 , 1897 .
1906 .
] Variations of Long Duration over Large Areas .
Archibald used data which extended up to the year 1893 and his deductions , which I think are the most recent , were:\#151 ; " The mean anomalies present all the characteristics of a true period , rising to a maximum of 0'0132 inch about the epoch of minimum sun-spot , and , with an exception in the sixth year , falling to a minimum of O'OIOO inch coincidently with that of maximum sun-spot , the former barometric epoch slightly preceding , and the latter slightly following the corresponding solar epoch as is usual in all other sun-spot comparisons . .
" Still the figures from the other years , and the repetition in each cycle , show that there is a cyclical tendency to high pressure at the time of few spots and low pressure at the time of many spots . . .
the amplitude of the variation amounts to 0'02 inch . .
By utilising in this inquiry observations made up to the most recent date possible , 1905 , it will be seen ( Plate 2 ) that the sun-spot maximum of 1893 corresponded with an epoch of mean low pressure about that epoch ; while up to 1901 , a year of about sun-spot minimum , the pressure had steadily risen .
It is thus evident that the same relationship is still in operation , only the amplitude is much smaller than was the case in the earlier years of observation So far as I am aware , the long barometric swing of 19 years in Australia has not been pointed out before ( Bruckner omitted the Australian area in his pressure investigation* * * S ) but the existence of a 19-year period of rainfall change has often been mentioned .
One of , if not the earliest record of it is in " Notes on the Climate of New South Wales , " 1870 , by Mr. H. C. Bussell , and in a later paper which he published in 1876 , f he more strongly advocated this rainfall cycle .
In a still later paper which he published in 1896 , Mr. Bussell collected information of a miscellaneous kind and extended his 19-year rainfall cycle both over a greater period of time and a wider area .
In fact Australia , India , Europe , Asia , Africa , North and South America , all tended to give him general ideas relating to droughts , which he marshalled and from which he deduced that this cycle was occurring over the whole earth , epoch for epoch , nearly simultaneously .
Mr. Bussell was finally led to give up the idea of the sun-spot cycle as influencing rainfall .
In place of solar action he regarded the moon as the origin of this cycle , a suggestion which he made in 1870S and again in 1896.|j * ' Klimaschwankungen seit 1700 , ' Eduard Bruckner , p. 194 , 1890 .
+ 'Journal of the Royal Society of New South Wales , ' vol. 10 , p. 151 , 1876 .
t 'Journal of the Royal Society of New South Wales , ' vol. 30 , p. 70 , 1896 .
S ' Notes on the Climate of New South Wales , ' 1870 .
|| 'Journal of the Royal Society of New South Wales , ' vol. 30 , p. 90 , 1896 .
Dr. W. J. S. Lockyer .
Barometric [ May 8 , It may , in conclusion , be stated that while the present investigation fully endorses the results of Indian meteorologists as regards the relationship between pressure and sun-spot variation , and agrees with regard to length of period with Mr. Russell 's long period cycle in Australia ( so far as the observations since 1857 permit ) , it does not corroborate his view that this cycle is world-wide in extent , with similar epochs of occurrence of the maxima and minima , or is dependent necessarily on lunar influence .
On the completion of this paper I took the opportunity of submitting it to Dr. W. 1ST .
Shaw , F.R.S. , and I should like here to express my thanks to him for kindly reading it through and making valuable suggestions .
Owing to the shortness of the time covered by the observations discussed in this paper I refrained from entering into any of the possible barometric changes which were visible in the curves other than those to which mention has already been made .
Dr. Shaw suggested , however , that as I had eliminated the variation of about four years ' duration to render apparent the changes of 11 and 19 years , I should proceed a step further and eliminate the 11-year change from the curves , since both the Indian and Australian pressure values exhibited changes of about thi3 length .
The object of this procedure was to find out whether the 19-year variation was a possible harmonic of the 4- and 11-year changes or a separate variation independent of these two shorter oscillations .
If the former , then it should be absent after the observations had been freed from these two variations .
To accomplish this , 11-year mean values were formed , that is , the value for each year plotted was composed of the combined mean for that year with the addition of the five years each side of this date .
Such an operation not only eliminates any periodic change of 11 years , but at the same time disposes of the variation covering about 4 years .
Plate 5 shows the resulting curves for India , East Indies , and Australia after the above-mentioned treatment .
The South American would have been similarly handled if the series of observations at each station ( excepting Santiago ) had been longer .
An examination of these curves shows that distinct variations still remain , and they are most marked in the Australian area , being specially prominent in the curve for Sydney and a little less so in that for Melbourne .
The variation at Adelaide is almost as strong as at Sydney , but , as before in the case of the 11-year change , is not quite of the same nature .
At Bombay a variation of much longer duration seems to be in operation , while the curve for Madras suggests maxima about 19 years apart , which in this respect conform with those of the Sydney curve ; the amplitude , however , is very 4- ana n-year Variations .
Cn 4- and 11-year Variations .
Oi 1860 0 5 18700 P -O * , l00U , y 5 l8buu 5 IB/ O 5 18800 5 18900 s 19000 hs Lurves to illustrate the Pressure Changes in India and Australia after the Elimination of the | 4- and 11-year Variations .
o* 18600 s 18700 Curves to ittust'1 ate \900'0 .
18900 5 , m/ U .
18800 \#174 ; Elimination of 5 '* India and Austraha Changes tn I\#171 ; * ( { ons .
C-V- \lt ; s\gt ; uyouu tM o c* :S Curves to illustrate the Pressure Changes in India and Australia after the Elimination of the 4- and 11 -yearVariations .
Roy .
Soc. ProcA .
vol. 78 , Plate 1906 .
] Variations of Long Duration over Large Areas .
much less .
The observations at Calcutta have been treated in a similar manner , but the resultant curve seems to represent a variation which differs from both those of Bombay or Madras .
The time dealt with in these curves is , however , so short that any reference to the possible periodic nature of these curves for the present is out of the question .
The variations exhibited are , however , very suggestive , and when further observations have been accumulated , a more definite idea of these barometric changes will be gleaned .
I take this opportunity to express my thanks to the Director of the Solar Physics Observatory , Sir Norman Lockyer , for permitting me to prepare this paper for the Society .
I am indebted also to Messrs. W. Moss and T. F. Connolly , computers in the observatory , for the abstraction , reduction , and drawing of the curves of the observations .
Conclusions .
Summing up the results arrived at regarding the behaviour of the curves representing these long barometric changes , it may be stated:\#151 ; 1 .
The Indian variation is nearly the inverse of the 11-year ( about ) sun-spot cycle , that is , years of mean high pressure are generally those of small spotted area .
2 .
The Australian variation is allied to , but in part a modification of , the Indian variation .
3 .
The interval between the Australian 'princmaxima is about 19 years .
4 .
The interval between the South American principal maxima is about 19 years .
5 .
The South American variation is not the inverse of that of Australia , but there is a time-phase difference between the maxima of about six years , the Australian maxima preceding those of South America .
6 .
There seems evidence to suggest that the 19-year variation is due to solar action modified by some terrestrial cause .
7 .
So far as the research has gone , no explanation has been found of the cause of the modification of the prominent 11-year variation in India into those of the pronounced 19-year changes which occur both in the Australian and South American regions .
[ Addendum , June 8 , 1906.\#151 ; Since this communication was sent to the Royal Society I find a still earlier reference to the 19-year cycle in Australia .
In an article on the " Development of Meteorology in Australia , " by Mr. Andrew Noble , of the Sydney Observatory , it is stated\#151 ; " Australian meteorology is greatly indebted to the Rev. W. B. Clarke for his untiring 60 Prof. G. Quincke .
Transition from the Liquid to [ June 6 , efforts in its behalf during those early years , beginning with his observations at Paramatta in the year 1839 and continuing long after the inauguration of the New South Wales service under Government auspices in the year 1858 . . . .
The 19-year cycle theory , elaborated by Mr. Russell in more recent years , was advanced by Mr. Clarke in the * Sydney Morning Herald ' of May 1 , 1846."* This reference is of great interest , since it indicates that this 19-year variation was evidently quite a prominent feature of Australian weather before the observations discussed in the present communication were made .
] The Transition from the Liquid to the Solid State and the Foam-Structure of Matter .
By G. Quincke , For .
Mem. R.S. , Professor of Physics in the University of Heidelberg .
( Received June 6 , \#151 ; Read June 21 , 1906 .
) 1 .
On June 19 , 1905 , I laid before the Royal Society the results of my researches on ice-formation and glacier-grains .
The further prosecution of these researches has shown that phenomena similar to those observed in the freezing of water occur in all bodies in nature , and are in agreement with the structure of metals as observed by myself and also by other investigators , who in some cases have described their results in the ' Philosophical Transactions .
' Solid bodies , then , are never homogeneous , but always exhibit a foam-structure .
2 .
For brevity 's sake I use the word " oily " to describe any liquid that tends to make its surface as small as possible , and exhibits surface tension at the boundary separating it from another liquid .
All liquids in nature resemble water in forming , as they cool , oily foam walls , which may be very thin and invisible .
The shape and position of these foam walls become visible on freezing or thawing in the following ways:\#151 ; ( a ) By fissures or fractures at the surface of the foam walls , whenever the liquid contents of the foam cells contracted on solidification , or when the walls and the contents of the foam cells contracted differently as they cooled .
* 'Monthly Weather Review , ' vol. 33 , No. 11 , November , 1905 , p. 480 .
Washington , U.S.A. , Weather Bureau .

