On the Surface Temperature of North Sea and Xorfh Atlantic. 11 



together our data, over continuous lines in space or continuous 

 periods in time, the better shall we discern the (jencral trend of our 

 phenomena, and the better and more ettectively will each one of our 

 numerical values be checked, tested, and interpreted by the conjoint 

 testimony of all its neighbours. 



Let us consider for a moment the various phenomena which our 

 subject involves, and the various kinds of diagrams which we may 

 draw in illustration of them. We are dealing with Temperature, in 

 relation to Space, and in relation to Time; and Space is for our 

 present purpose si)ace of two dimensions only, for we are only dealing 

 with the temperatures of the surface, and not those of the depths of 

 the sea. We have therefore four "dimensions" to deal witli, viz. 

 (1) Temperature, (2) Time, and (3,4) Space in latitude and longitude, 

 or in any other two correlated directions that we please. Now any 

 system of diagrammatic representation that we can make upon a 

 sheet of paper is practically confined to two or tq at most three 

 dimensions, the two variables corresponding to the ordinates and 

 abscisste of an ordinary diagram on which we plot a curve, and the 

 third variable being introduced when, upon our system of co- 

 ordinates, we superimpose a system of " contour-lines " or " isopleths." 

 We must use, accordingly, more than one system of diagrams before 

 we can represent and correlate the phenomena of four dimensions 

 with which we have actually to deal. 



The chief types of diagram which suggest themselves as useful, 

 whether tw^o-dimensional or three-dimensional, are the following. 



Our two-dimensional diagrams are meanwhile limited to those in 

 which {a) we plot Temperature against Time, or (6) Temperature 

 against some one dimension in space ; for instance, when (plotting 

 temperature against time) we draw the annual curve of temperature, 

 say, at Dundee ; or when (plotting temperature against space or 

 distance) we draw^ the curve of mean temperature between Leith and 

 Hamburg. 



These two types of two-dimensional diagram we may convert into 

 three-diniensi( »nal diagrams, by superimposing additional Temperature 

 curves upon the one already drawn ; for instance {<t) by using our 

 first diagram for the seasonal tiuctuations at several stations, e.g. 

 Aberdeen, Dundee, Leith, etc. ; or (l) by introducing the element of 

 time into our second diagram, and fitting into it separate temperature 

 curves for the individual months. 



(c) The commonest of our three-dimensional diagrams is that in 

 which we first plot latitude against longitude in space, or in other 

 words, draw an ordinary map or chart of a portion of the earth's 

 surface: and then plot, over this two-dimensional diagram, the 

 variations of temperature, in the form of isotherms or contour-lines. 

 This method of representation, first introduced by Humboldt, gives us 

 what we ordinarily call a " temperature chart " ; but observe that this 

 temperature chart is, and can be, only for one epoch of time. It may 

 be for the average temperatures of the year, or it may be for the 

 mean of some particular month ; but it cannot show the succession 

 of phenomena from month to month. To show this, that is to say, 

 to introduce a variation in time, we must sacrifice one of our other 

 dimensions. We cannot here sacrifice Temperature, for that is the 



