It is advisable to take several observations, and the safest way is to take one first with the index glass uppermost, and then with the instrument inverted, after which to return to the first, and so on for two or three times each way. In the pages which follow, there is given a table containing the result of all the observations made during this voyage, preceded by several sets of observations in the fullest detail. From the table it will be observed how seldom the dip, actually measured, agrees with that inferred from the mean refraction. Some of these experiments shew very remarkable differences, and point out the great utility of this instrument. The practical navigator, particularly if he has been in hot climates, will recollect how discordant his observations for latitude always were, and how few even of the best observers agree in their determination of the latitude of the same place, simple as the observation is thought to be. The cause is quite clear; and though it equally affects altitudes taken for absolute time, the disagreement is less obvious, and it will often happen that a chronometer going extremely well appears to vary every day from inaccuracy in the observations. Thus it is, I think, generally admitted, that it is almost impossible to rate a chronometer from altitudes observed with the sea horizon. Nor is this difficulty removed by taking equal altitudes, because the refraction in all probability will be different at the two observations. With an artificial horizon, indeed, the changes in refraction are not felt, because, at a considerable elevation above the horizon, the changes are very trifling. But it often happens in practice, that the artificial horizon cannot be used, and we are then reduced to the sea horizon, where the changes of refraction are always the greatest. In the Yellow Sea, for instance, we had no opportunity of landing during all the time that the squadron was at anchor, till the day before we sailed. So that during nearly a fortnight that the ships were at anchor, the sea horizon was necessarily used. I need only to refer to the observations taken off the Pei-ho, viz. from No. 37 to 62, to shew how extremely fallacious the results must have been. It is much to be wished that this excellent instrument should be brought into general use in navigation. THE FOLLOWING EIGHT OBSERVATIONS ARE SET DOWN IN THE FULLEST DETAIL, IN ORDER TO SHEW THE METHOD USED IN RECORDING THEM. No. 31. YELLOW SEA. _July 23, 1816._--6 P.M. Index uppermost. Instrument inverted.