The general appearance of the instrument is well shown in the figure, which is engraved from a photograph I took of Mr. Ekholm while actually engaged in talking through a telephone to M. Hagstroem as to what portion of a cloud should be observed. The latticework tube, the cross wires in place of an object glass, and the vertical circle are very obvious, while the horizontal circle is so much end on that it can scarcely be recognized except by the tangent screw which is seen near the lower telephone. Two such instruments are placed at the opposite extremities of a suitable base. The new base at Upsala has a length of 4,272 feet; the old one was about half the length. The result of the change has been that the mean error of a single determination of the highest clouds has been reduced from 9 to a little more than 3 per cent. of the actual height. At the same time the difficulty of identifying a particular spot on a low cloud is considerably increased. A wire is laid between the two ends of the base, and each observer is provided with two telephones--one for speaking, the other for listening. When an observation is to be taken, the conversation goes on somewhat as follows: First observer, who takes the lead--"Do you see a patch of cloud away down west?" "Yes." "Can you make out a well-marked point on the leading edge?" "Yes." "Well, then; now." At this signal both observers put down their telephones, which have hitherto engaged both their hands, begin to count fifteen seconds, and adjust their instruments to the point of cloud agreed on. At the fifteenth second they stop, read the various arcs, and the operation is complete. But when the angles have been measured the height has to be calculated, and also the horizontal and vertical velocities of the cloud by combining the position and height at two successive measurements at a short interval. There are already well-known trigonometrical formulae for calculating all these elements, if all the observations are good; but at Upsala they do far more. Not only are the observations first controlled by forming an equation to express the condition that the two lines of sight from either end of the base should meet in a point, if the angles have been correctly measured and all bad sets rejected; but the mean errors of the rectangular co-ordinates are calculated by the method of least squares. [Illustration: N. EKHOLM MEASURING CLOUDS. This figure shows the peculiar ocular part of the altazimuth, with the vertical and horizontal circles. It also shows the telephonic arrangement.] The whole of the calculations are combined into a series of formulae which are necessarily complicated, and even by using logarithms of addition and subtraction and one or two subsidiary tables--such as for log. sin squared([theta]/2) specially constructed for this work--the computation of each set of observations takes about twenty minutes.