TRIGONOMETRICAL SURVEY OF NEW SOUTH WALES. 221 
advisable to follow the plan urged by General Walker* and 
Dr. Gill.t The latter uses the following words :— 
“The method of employing groups of neighbouring astrono- 
mical stations instead of isolated stations is unquestionably the 
best plan for dealing with the difficulty. Having regard to the 
fact that the probable discordance produced by deviation of 
the plumb line is far greater than the probable error of an astro- 
nomical latitude as determined by observations of a single night, 
it must be always more advantageous to observe for latitude on a 
single night at each of five or six stations in the neighbourhood 
of a principal point than to make a long series of observations at 
any one point for the purpose of securing an accuracy in the 
determination which is in great part nullified by unknown local 
attraction.” 
“And if this be true for meridian arcs, as General Walker’s 
results (loc. cit.) have abundantly proved to be the case, it must 
be equally true for longitude stations. Every principal longitude 
station should therefore be surrounded by neighbouring stations 
for the purpose of eliminating, or at the least diminishing the 
effect of local attraction. The establishment of a longitude station 
is an incomparably more laborious and costly matter than is that 
of a latitude station ; it involves, for a like accuracy, a larger 
number of observing nights, an exchange of instruments and 
observers, and the cost probably of special wires, thus the labour 
and cost render the method practically prohibitory.” 
But the same end can be accomplished in a much simpler and 
more accurate manner by means of azimuths. Whatever informa- 
tion as to the amount and direction of local attraction can be 
derived from observations of longitude, the same information can 
also be derived from azimuths. If each principal longitude 
station was surrounded by six symmetrically placed astronomical 
stations, the lines joining the stations would form a regular 
hexagon with a central point—a figure which is the most favour- 
able possible for accurate geodetic measurement. If the 
astronomical latitudes and longitudes of these points are then 
determined, we have, from a discussion of the discordances 
between the geodetic and astronomical results of the figure, all 
the requisite data for computing the local attraction at the central 
point, or rather, we secure all the advantages which would result 
from a group of seven latitude and longitude stations. This 
method would be entirely free from the objection which can fairly 
be brought against the use of azimuths as a substitute for longitude 
operations, viz., the accumulation of error which is inevitable in 
long chains of triangulation. 
* India’s Contributions to Geodesy. by General J. T. Walker, R.E., C.B., &c. Phil. 
Trans., Vol. 186, pp. 778-787. 
+ Report on Geodetic Survey of South Africa, 1896, already cited. 
