REPORT ON A LEVEL LINK. 15 



A rigid adherence to this sj'stem rendered it improbable that 

 a wrong reading could be written down, without immediate de- 

 tection : — in fact, such an instance does not appear to have oc- 

 curred. Had it even been so, a discrepancy must have existed 

 between the columns of different scales, which would have been 

 readily detected on casting up and compai-ing the totals, at the 

 end of the day. From erroneous readings, therefore, it is evi- 

 dent, there was little or nothing to fear ; but these are far from 

 being the only, or the principal sources of error. On one or 

 two occasions, we were very near committing a mistake, in be- 

 ginning at a different station from the one on which we had 

 previously closed. This would have occasioned an error, per- 

 haps of large amount, which could only have been detected by 

 the second and independe3it series of levels, taken over the 

 ground in an opposite direction. For this i-eason alone, I should 

 not consider it safe to depend on one course of levels only, 

 whatever may have been the precautions used to guard against 

 error. 



The total length of my line of leveling between Portishead 

 and Axmouth, besides the branch lines to Bristol and East 

 Quantockshead, is about 74 miles. This distance was divided 

 into separate stages ; each of which, averaging about 10 miles 

 in length, was twice leveled over, first in one direction, and 

 then in the opposite, before the next stage was commenced. It 

 is very remarkable, that with a few partial exceptions, the heights 

 of all the points touched upon by both series, came out less by 

 the levels returning, than by the levels going : so that the first 

 station,or starting-point, always appearedlower when Ireturned, 

 than it was at my setting out. But as the height of this point 

 is the same in both cases, the error must, of course, be thrown 

 on the distant point, or station at which the returning levels 

 commenced, which reverses the first apparent differences, and 

 makes all the heights in the second series progressively greater 

 than those in the first, the most distant point having the greatest 

 error. The following table gives the differences thus found at 

 20 points along the line between Portishead and Axmouth, the 

 height, in every instance, coming out greater from the series of 

 levels returning towards Portishead. 



