III .
" A Tabular Form of Analysis , to aid in tracing the Possible IInfluence of Past and Present upon future states of Weather .
" By S. ELLIOTT HOSKINS , F.R.S. , &amp;c. Received March 28 , 1867 .
The data upon which the present communication is founded are derived from the ' Greenlwich Reports , ' from Mr. Glaisher 's papers in the Philosophical Transactions , and from my own observations at Guernsey .
The latter were commenced in the autumn of 1842 , in accordance with the recommendations of the Committee of Physics of the Royal Society , and were taken at the request of Professor Daniell , by whom the irnstruments employed were selected .
These instruments , made by Newman , were after a time replaced by others , at the suggestion of Mr. Glaisher , by whom they were compared with the standards at the Royal Observatory .
For my own guidance in the first instance , I sought to arrange the results thus obtained in such a marnner as to discover , if possible , whether any month or class of moniths stood to each other in the relation of cause and effect ; in other words , whether the atmospheric conditions of autumn exercised any distinguishable influence upon the fruitful or unfruitful character of ensuing seasolns .
In order to attain this object , the priilciple seemed to be that of condensing within narrow limits , by means of intelligible symbols , as many elements of weather , in the popular acceptance of the word , as might be required .
But the ordinary curvilinear form of diagram could not be so modified as to answer this purpose , and I therefore availed myself of a plan suggested by Mr. Galton:-that of converting the records of observations into appropriate signs , and placing them compactly in a series of squares .
Upon this principle the annexed diagrams* are constructed , comprising those elements of weather which more directly affect vegetation ; viz. heat , cold , dryness , moisture , and their combinations .
The same kind of preparatory steps were taken for the compilation of the Greenwich as for the Guernsey diagram , so as to render the results comparable-less , perhaps , for the sake of mere comparison , than for the purpose of testing the value of the latter by means of an accredited standard .
The first process consisted in copying out the degrees of monthly mean temperature , the number of rainy days , and the days of wind , from four directions , intermediate to the cardinal points .
These several copies being verified , the monthly average of each of the above elements for twenty years , from 1843 to 1862 , was taken .
The next step was to obtain the difference between the adopted average and the mean of each month in every year .
By prefixing the plus and minus signs to the resulting figures , the excess and defect of each element is shown .
The third process was to separate the above-mentiorned series of years into two decennial periods or decades , and then to compute , not only the number of months above and below the average , but also the degrees of temperature .
( See Tables I. , II .
, III .
, V. and VI .
in the Appendix .
) Lastly , the numerals thence derived were converted into .
simple and familiar signs , which were then delineated upon a sheet of sectional paper , accurately engraved according to scale .
A square space was allotted to each of the months , and they were laid down as abscissas , with the years for ordinates .
( See Diagrams I. and II .
( Archives ) .
) The light squares in the diagramis deniote warmth , the degrees being expressed by a modification of the plus sign , in red inik ; the dark , or shaded squares indicate cold , and the degrees are marked in black ink with the minus sign .
Black dots indicate the number of rainy days above the average , and red dots the number below it .
The diagonlal linies from the centre of each square show the direction and days of predominant wind , one-sixteenth of an inch being equivalent , to five days beyond the mean ; the red lines are associated withi dryness , and the black with moisture .
It may be necessary to explain that , as regards rainfall , frequency , rather than quantity , was selected as a criterion .
That is to say , the number of days on which rain fell rather than the number of inches collected by the pluviometer ; for it not unfrequently happens that a few heavy showers yield a greater amount of wa.ter than many days of gentle rain of long , continuance .
The combined signs in the above arralngement are intended to represent four different states of weather , viz. warm+dry ; warm+wet ; cold+dry ; and cold+wet .
If the sectional paper be large enough to admit of blank spaces being left , the signs of the weather may be delineated therein as each mouth elapses ; and thus the diagram becomes a sort of register , and an ever-ready table of reference .
For instance , by rulning the eye along the vertical columns of the Guernsey diagram , to which I must confine myself for the present , the existing state of the weather can easily be compared with that of corresponding months , up to 1843 ; and by following each horizontal line of squares the character of each year may as readily be ascertained .
Thus we shall find , on comparing the September of 1865 with that of preceding years , that it was the hottest and the driest of the whole series .
On looking along the ordiniate corresponding to 1846 , it will be seen that during , eleven months of that year the temperature was uniformly above the adopted average .
On taking a general view of this diagram , after its completioni , a vervy cursory glance sufficed to show me that a striking differenice existed in the distribution of lig , ht and shade .
On closer inspection , it became maniifest that the number of light squares in the one decade exactly couniterbalanced the dark squares in the other ; so that the warm months of the first period were in direct ratio to the coldI months of the second .
These contrasts , quite unlooked for by me , were all the more surprising , as the data employed had been taken as they came , and not selected for the purpose of supporting any preconceived notion .
It seemed to me , therefore , that this kind of diagram , besides serving as a convenient table of reference , was a collection of materials prepared and classified for further analysis .
Under this impressiou I proceeded to decompose it , and to rearrange the products in a tabular form*-converting into letters of the alphabet the combine ed signs in the squares , so as to designate the foir states of weather , before mentionied , as follows:-A =war'a ' + dry ; B= warm + wet ; C= cold + dr , and D= cold + wet .
These letters were them placed in columns under the heads of mnoniths and years ; the number of timtes in which each letter recurred was noted , and these numerals , which may be termed coeecients of the sums of the letters , were collected in linles and columns , those of the months at the foot , and those of -he years at the sides of a table of atiialsis .
See Table IV .
When the coefficients of the whole series were thus placed in .
juxtaposition , it was satisfactory to find that the general contrasts , noticed in the diagram , were bornie out numerically ; and still more satisfactory to ascertain that there was a close agreemenit betweeni the ratio of the months and that of the degrees of temperature , plus and minus .
Months .
Degrees .
First decade .
78 warm to 42 cold .
1530'5 pltus to 550 2 imiinUs .
G er Second decacae .
78 cold to 42 warm .
1530.0 mntus to 000.0 plus .
e nacy The columins at th e sides ol the Tables of anilysis , that of Greenwich as well as Guernsey , indicate that there was the it trtrsion of one cold year ( 1845 ) in the warm perioc , and of two warm years ( 1857 and 1859 ) ill the cold period .
A similar kind of inltercalation was pointed out by Mr. Howard , in his I Cycle of the Seasons , ' from 1824 to 184 1 , i amely , the intrusioni of one cold year in the warm , and one warm year in the cold cycle .
On examinint n the coefficients riore in detail , in the hope of being -able to detect some group of inoiiths which seemed to bear a peculiar relation to the rest , I met with a frequent recurrence of anx exact inverse order between the warm months of the two decades ; and oft a direct ratio between the wet and dry .
Durung the first decade , the ratio between wart and cold , in the groups of January , February , March , April , August , Septemnber , October , and December , is invariably 6 to 4 .
It is one of greater inequality in the Junes and Novehbers , being 8 to 2 in both cases ; warm MJays are equal to the cold , but warm Julys preponderate in the proportion of 9 to 1 .
In the second decade , the rarm Novetlbers are to the cold the exact reverse of whiat tiev were in the first , being 2 to 8 ; and the ratios of Ja* I arn indebted to Dr. Gtuy 's Croonian Lectiires for n insiglt into this method .
nuary , February , iarch , and April are also reversed , being 4 to 6 , instead of 6 to 4 .
Wbhen onie decade is compared with the other , the last-named months are found to stand in exact inverse order , viz. 6 to 4 in the first , and 4 to 6 in the second .
The wet and dry groups ceern , on the whole , to be more evenly balanced than the warm and cold ; but the Novernbers of the second decade are remarkable for dryn-ess .
The connexion between the predominant winds and the other states of weather has not as yet been traced systemiatically but the diagram shows great excess of north-east wiid in thespring 01 18552 , and a loog conitin uanee of cold weather sCttiDg in early in the following year .
It is also evident that wind from south-eastwas nor prevalent ( uining thie second thanthe first decade .
Prom the foregoing comparisoni of the different months , the groulp of Novembers seems to be the most exceptional ; it may therefore be worth wshile to recapitulate the peculiarities that have been nioticed .
Lsr .
Th-e r'atio between the coefficients of this group in di ; erenit decades is invariably onie of con siderable inequality .
2nld .
Two cold Novembers only occur in the warm cycle , and only two w1arm ones in the cold cycle .
3rd .
In the second decade the proportion of warm to cold Novembers is 8 to 2 , and of dry and wet 2 to 8 ; but in the warm period warm and wet m11onths were prettily evenly distributed .
4tl .
ioviembers of conmparatively low tenmperature , such for instance as those of 1851 , 1853 , 1854 , 1855 , 1856 , 1860 , 1861 , and 1862 , were in eaeb year succeeding those enumerated , followed by Ma is or Junes of a sirmilar character .
The following rs me ' shows the relations between the Novembers and the Junes .
M-onths .
Nove , mbers .
Jtnes .
Gue r Isey .
1st decade .
8 warmii to 2 colcl .
8 warmrL to 2 cold .
9i 2n decade .
8 cold to 2 warim .
7 cold to 3 warm .
DIegrees .
Novembers .
Jllmes .
Iue st n esade 15{7 %plus to 40^2 minus .
17 ' ?
5 phlis to 3'-O m-tinus .
luernsey- .
2nd deladl.e .
14 ?
` 0 mi-nus to 23-8 plus .
180 5 minus to 3 ?
.1 plus .
These contrasts and analogies seemi to ju1stify the surmise that the atmospheric conditions of the former monthis may have exercised some irifluence upon those of the latter .
WAThether such be the case generally is not to be determinied until a miiuch longer series of results at Guernsey can be compared .
The peculiarities withi respect to the Novembers may he purely accidental , or confined to the period under consideration ; but that they are 2 it jot restricted to locality is proved by the Greeniwich Tables , in which these groups staned out still more prominenitly ( see Table VII .
) ; the ratios between warm and cold being 9 to 1 in the first decade , and 1 to 9 in the second .
It is difficult therefore to avoid the conclusion that , during the twenty years in question , the Nove-n-bers were exceptional months at both places ; although at Guernsey they were more frequently followed by unfavourable Junes .
The Greenwich diagram ( Diagram II .
Archives ) , to which I must now briefly advert , does not exhibit so strikin , a contrast of light and shade as was observable at first sight in the other diagram .
B1ut on fri-ther examination it will be found that the warm months of the first decade correspond nearly in number with the cold months of the second , although not so exactly as at Guernsey .
Months .
Degrees .
Greenwch .
f Ist decade .
65 warnm to 55 cold .
157 ?
-5 plus to 1080.5 imiinuis .
Greenwil .
* 2nd decade .
62 co]cl to 58 warm .
148 ?
4 minus to 11 6 ?
8 plus .
Oni comparing the above abstract with that in a previouis page , it will be perceived that the disparity between the general results , from both places , is not very considerable ; a similarity all the more remarkable , whenl we consider the great difference in position and latitude of the inlland and the insular stations .
See Diagram III .
It would be superfluous to enter into any further discussion of the various alterniations which the coefficienlts are susceptible of , in a paper which is merely intended to direct attention to the accompanying diagrams and arnalytical tables .
My motive for venturing to bring them under notice is a desire to place them in the hands of those better qualified than I am for conducting processes of iniduction ; and as the modified plan I have adopted is based upon long recog , nized principles , which are applicable to the investigation of atmospheric phenomena in any locality , I trust that it may be deemed worthy of consideration .

