still but imperfectly, from the influence of the irregular anomalies; but they are evidently still charged with the secular change. In order to eliminate this, its amount between the middle of each month and the 1st of October must be applied with the negative sign for the first six months, and with the positive sign for the last six months. Assuming the above determined value for the twelve months, 3′ 46″·2, we obtain the following results.
| 8, a. m. | 1, p.m. | Mean. | |
| April | − 2 35·6 | + 0 54·2 | − 0 50·7 |
| May | − 1 55·3 | + 1 10·0 | − 0 22·6 |
| June | − 1 6·6 | + 0 56·7 | − 0 4·9 |
| July | − 0 32·0 | + 1 13·6 | + 0 20·8 |
| August | − 0 18·8 | + 2 20·7 | + 1 0·9 |
| September | − 0 43·0 | + 0 41·6 | − 0 0·7 |
| October | + 0 9·3 | − 0 11·0 | − 0 0·8 |
| November | + 2 8·0 | − 1 24·7 | + 0 21·6 |
| December | + 2 13·3 | −3 9·0 | − 0 27·8 |
| January | + 1 50·3 | − 1 51·5 | − 0 0·6 |
| February | + 1 21·3 | − 1 40·1 | − 0 9·4 |
| March | − 0 30·9 | + 0 59·6 | + 0 14·3 |
In these results we find as much regularity as could be expected from the observations of three years only. The first column shows how much the forenoon declination in each month differs from the mean forenoon declination, and in the same way the second column shows the difference of the afternoon declination in each month and the mean afternoon declination. It must be recollected that the latter mean itself is 10′ 23″·8 greater than the former mean.
It appears remarkable that in all the twelve months the forenoon and afternoon declinations fluctuate about their mean values in opposite directions. In the five winter months, from October to February, the forenoon declination is greater, and the afternoon declination less, than their respective mean values; and both circumstances tend, during this portion of the year, to render the differences less than their mean amount. In the other seven months exactly the opposite effect takes place. Moreover, these opposite fluctuations are, upon an average, nearly of equal magnitude; the consequence of which is, that they nearly destroy each other in the mean given in the last column. This result may be thus stated in other words: the mean
