remembered, under the most favourable circumstances. From this we may see that a single set of lunar distances, even when taken by an experienced observer, is of comparatively little value. But here again the art of observing comes in, and if several sets be taken east and west of the moon, the errors produced on one side have a contrary effect to those taken on the other, and thus in the mean results they neutralise one another. As an instance of this, I may mention that Consul O'Neill fixed the longitude of Blantyre, to be used as a secondary meridian by several hundred sets of lunars, for which he received the Gold Medal of the Royal Geographical Society. Few travellers, however, possess Mr. O'Neill's skill, and fewer still, I fancy, can be found who are prepared night after night in a sickly climate, to spend the time, when others are asleep, in measuring lunar distances. All that I can say in their favour is this, that the opportunities for observing them are of more frequent occurrence than for occultations, moon culminating stars, or the eclipses of Jupiter's satellites. But they require far more skill to take, and, as a whole, can hardly be said to give satisfactory results, unless taken by a practised hand who thoroughly understands how to balance his observations.

There remains one observation of considerable importance, which is that for finding the error of the compass. This observation is very much like that which I have previously described for finding the time, and, by aid of the diagram (Fig. 3), I think I can explain to you how we find the true bearing of a heavenly body.

We observe its altitude above the horizon, and, in the same manner as before, we subtract it from ninety degrees to get the zenith distance. Then, by a well-known formula, we compute the angle at the zenith, which will be equal to the distance in degrees of the heavenly body from the meridian, which, being north or south, will at once give us the true bearing of the heavenly body. Having previously taken the bearing of this body with our prismatic compass, by comparing the compass bearing with the true bearing, thus found, we are at once enabled to ascertain its error. This observation is of considerable importance, as all bearings taken with a compass require to be corrected for this error before the true bearing can be obtained.

We will now pass on from astronomical observations to those that are taken for fixing detail in a map. I will first speak of the manner in which the heights of certain features in a country are fixed. The most accurate of these is by triangulation, but as it hardly comes within my province to enter into the details of a trigonometrical survey, I will therefore pass on to the method of finding the difference of height by means of the mercurial barometer.

Now in the first place there is considerable difficulty in conveying a barometer filled, from one station to another, as, owing the vacuum in the tube, the mercury when set in motion begins to pump, and, more frequently than not, knocks the end out of the tube. Many attempts have been made to get over this difficulty, and we have several barometers now which can be carried with comparative safety, even in a rough country. The first of these that I will mention is George's barometer.

In this case we carry empty tubes, and fill them by a cool process when we arrive at the point where the observations are to be taken. In temperate climates and at moderate elevations this can very conveniently be done, but at the top of a mountain, standing among the ice and snow, it is a very trying operation.

There is then another barometer, the Boileau-Mariotti, which is an extremely portable instrument, requiring no filling, and giving very excellent results.

The manner in which the observations for height are taken with a barometer is as follows: A reading of the barometer is taken at the lower station, and also the temperature of the air in the shade. This is repeated again at the upper station, the elevation of which we desire to discover. With these two readings and the mean of the temperatures, we are able to ascertain, with approximate accuracy, the difference of level between the two stations.

The aneroid is also very generally used for this purpose, and the observation is precisely the same as that taken by the mercurial barometer. It is, however, liable to mechanical disturbance, and cannot be relied upon for heights above 10,000 feet. A very good plan when observing heights with aneroids is to take two, one of which should be so constructed that it only commences to work at the point where the other leaves off. Wherever the aneroid is used, it should always be checked by the boiling-point thermometer, by which means any great error that may have been caused by a blow or a jerk will be at once perceived. The boiling-point thermometer itself is a very useful little instrument for determining heights. It is not, however, to be relied upon within 100 feet. This I have put to the test on many occasions. Once I carried on a series of observations with a boiling-point thermometer at two distant stations with a scientific friend. We both had Greenwich time and took our observations simultaneously, and we found that, after a fortnight's daily observations, there had been a variation of nearly 100 feet from the true height of the upper station. However, at great elevations 100 feet does not matter very much, especially as the height would only be determined by a boiling-point thermometer in a preliminary survey.

I am now going to introduce to your notice two instruments which are peculiarly adapted to the wants of travellers, and the use of which could be very easily taught in schools: these are the plane table and prismatic compass. The plane table is simply a piece of wood on which paper is stretched, mounted on a tripod. The alidade is fitted with two sight vanes, and the instrument is also furnished with a long compass needle in a box. Now, by referring to the diagram (Fig. 5), you will see how this is used.

In the use of the plane table it is absolutely necessary, when the table is set up at any station, that it should be oriented-that is to say, its sides must be parallel to the position which the same side of the table had on a previous station. This may be done approximately by the compass needle, but where possible it should be oriented by placing the edge of the alidade on one or more rays that have been drawn on the paper, and turning the table until the sights come on with the object itself.

You may also

and thus all considerable errors are detected at once. see that being simply a piece of board mounted on a strong tripod, any

The great advantage of using the plane table is that the map is actually constructed on the board while the work of surveying is going on,

Fig. 5.-Illustrating the manner in which the Plane Table is used.

part of it can easily be repaired, should it become broken or temporarily disabled.

The prismatic compass is the most portable of all surveying instruments. The manner in which the survey is made is similar to that of the plane table, the only exception being that the results are taken down in a note-book, instead of being drawn, as in the case of the plane table, on the paper in the field.

We have a diagram here (Fig. 6) which illustrates the manner in which a survey through a jungle can be plotted.

In this case you will see that a scale is given of so many minutes of time by the inch, and the reason for adopting this method is that in a rough country, or one where the traveller is constantly turning about in the jungle, it would be impossible for him to arrive at anything like an accurate estimate of the number of miles that he had travelled over. But, by taking the time occupied to travel from one point to the other, he at any rate keeps up a just proportion all through his survey. Mr. Holt Hallett made a journey of more than 1000 miles through parts of Burma, Siam, and the Shan States, during which he adopted this method, with remarkably accurate results.

A simple and rapid method of making a rough survey of a limited area is shown in the diagram on the next page (Fig. 7).

It consists of taking bearings with the compass, and measuring the distances with the Weldon range-finder. The manner in which this system is carried out is as follows: A flagstaff is driven into the ground in some convenient position, to which a surveying chain or measuring

tape is attached by a ring. From this point the bearing of the object is taken, and then, by means of the first prism of the range-finder, a staff is set up at a right angle to the line joining the flagstaff and the object; the observer then retires, keeping the two staves in a line until, in the second prism of the range-finder, the object and the flagstaff appear in one. He has then only to measure his distance to the flagstaff, and this, multiplied by fifty, will give him the distance of the object. The Weldon range-finder is no larger than an ordinary watch, and has three prisms; one is accurately constructed to 90° for setting out right angles, and the other two are 88° 51' 15" and 74° 53′ 15′′ respectively. Of

course with a sextant set to these angles the same work could be done, but the range-finder has the advantage of being much more portable, and not liable, as the sextant is, to get out of adjustment.

I will now, with your permission, make some remarks on the projections most suitable for travellers, for, although map projection does not come under the head of observing, it is nevertheless a kindred subject, as it is on a map that the results of such observations as I have mentioned are exhibited. For any place within the tropics Mercator's projection is the most suitable, chiefly from the fact that a straight line drawn through any part of it will intersect all the meridians at the same angle, thus showing correctly all the relative bearings.

If you will look at the diagrams on which the same line is plotted, you will see that in the case of the Mercator the track appears as a straight line, intersecting all the meridians at the same angle, while on the conical projection a straight line would intersect each meridian at a