36 THE VOYAGE OF H.M.S. CHALLENGER. 



Now, in an inquiry into the comparative mean distribution of atmospheric pressure, 

 it is clear that the first, and indeed, as respects time, the essential requisite, is that the 

 means be drawn from observations made in the same years. In tropical and most sub- 

 tropical regions, where the mean pressure differs but little for the same month from year 

 to year, that the observations be for the same years is not a matter of such paramount 

 importance ; but elsewhere, owing to the more or less marked instability which prevails 

 with regard to pressure, it becomes of the utmost importance to obtain the means of 

 observations for the same years. 



Mean Pressure. — The mode in which the observations were discussed was first to 

 extract, for each country by itself, the mean monthly pressures reduced to 32°, where 

 these were obtainable, year by year. Since in this way the curve of variation from 

 month to month was easily kept in mind, many typographical errors, faulty averages 

 calculated from portions of months only, and other anomalies, were detected, and 

 these doubtful means were at once inquired into and rectified. 



As the work advanced, the mean annual pressure, further reduced to sea level, for 

 each station for which observations for the whole of the fifteen years were available, 

 was entered on maps of the countries. The results for every country showed anomalies 

 and discordances in the barometric means, which called for inquiry with a view to their 

 rectification approximately. 



No inconsiderable number of errors were occasioned by incorrect heights. These 

 have been rectified by correspondence ; but in cases where no levelbng or trigonometrical 

 survey has been made, approximate heights have been adopted, deduced from the annual 

 chart of mean pressure. Some errors were found to be due to the state of the barometer, 

 or to its verticality. But the larger number of anomalies had their origin in the personal 

 errors of the observers, arising mainly from the different methods employed in setting 

 the vernier of their barometers. These may be classed as under : (1) setting the vernier 

 in the line of that part of the top of the mercury which is in immediate contact with the 

 glass tube, the instrument being thus read about - 033 inch too low, more or less, 

 according to the diameter of the tube ; (2) setting the vernier by bringing it down 

 till the speck of light on each side is on the point of disappearing, the error in this case 

 being from 0*008 inch to 0'020 inch too low, according to the breadth of the slit; 

 (3) setting the vernier so that a clear space is left between it and the tangent to the 

 mercurial curve, the error in this case being about O'OIO inch too high. The last 

 method of reading is mostly caused by weak or failing sight, the observer not being 

 aware that a lens or spectacles is now required, and consequently it does not materially 

 affect the observations, when two readings are made, as from a Fortin or siphon 

 barometer. It leads, however, to the above error of about O'OIO inch with Board of 

 Trade and other barometers, which take no account of the height of the mercury 



