( 707 ) 



If we suppose that during each of the periods of 18 vears the 

 mean sea level has remained unchanged we deri\ e from the deviations 

 of the yearly means from the mean of 18 years a mean error for 

 each year of ± 25,1 mm. and in the mean of 18 years a mean 

 error of ±5.9 ram. 



If on the contrary we suppose that during each oï the periods of 

 18 years the mean sea level with regard to the adopted Amsterdam 

 zero has varied proportioniilly to the time, we get for the mean 

 error of the yearly mean ± 24,3 mm. and for the yearly variations: 



from 1700—1717 +1,57 mm. +1,10 mm. 

 „ 1796—1813 -f0,14 ., +1,10 „ 

 „ 1843—1860 —2,30 „ ±1,10 „ 



Hence in the 1=^ and 2"'^ periods, in agreement with the general 

 variation of the mean sea levels from 1700 — 1826, the mean sea 

 level has apparently come neixrer to the adopted A. P., but has retired 

 thence in the 3'^ period in agreement with the variation from 1825 — 

 1851,5. Nevertheless the mean errors of each of these yearly variations, 

 + 1,10 mm., are so large with regard to the variations themselves, 

 that we attach only a very small weight to the values found ; only 

 to the yearly \ariation in the 3^ period, more than twice the value 

 of the mean error, we may attach a somewhat larger weight. If 

 we adopt a uniform yearly variation between the years 1708,5 and 

 1804,5, this would amount to 0.58 mm.: in good harmony with 

 this are the results for 1725 and 1749, but the result for 1775 

 shows a deviation of 32 mm. 



We conclude that the elevation of the adopted A. P. above the 

 mean sea level has gradually varied and that the \ariations can be 

 considered as partly proportional to time ; they cannot howe\ er 

 be derived exactly from the observations. 



The ele\aiion of the A. P. in the tidal station above the mean sea 

 level in the tirst and the last year of the series of observations, 1 700 

 and 1860, are according to table II 162 mm. and 75 mm. each 

 with a mean error of ± 25 mm. In order to obtain for these 

 elevation %alues with a smaller mean eri-or, we may use, upon suppo- 

 sition that no sudden variations have taken place in the zero of the 

 gauging rod, the elevation observed in closely preceding or following 

 years, which must be reduced to the year 1700 or to 1860 with an 

 adopted yearly variation. Because the yearly variation is not known with 

 great precision, as appeared above, it is desii-able that these years 

 should not be at a great distance from 1700 or from 1860 ; therefore 



48* 



