cll, G. R. CROWDERY. 
On 2nd December, 1893, the level of each rail at the joint and 
centres was carefully taken, and again on the 14th August, 1894. 
The greatest variation was :02 of a foot, the difference being only 
such as would be expected by rails seeking their proper bed on 
the sleeper. ‘ 
No alteration could be traced in the expansion allowance 
between the dates mentioned, and the rails have also remained in 
good line. It was thought that if there were a tendency for the 
rails to expand, their course would be directed to the point of 
least resistance, and it was believed that under the conditions it 
would be in an upward direction rather than laterally, hence the 
object of levelling. It is only necessary to add that after nearly 
four years’ constant and fairly heavy traffic, some difficulty is 
experienced in finding where the joints butt. The author must 
admit, however, that several joints show some slight signs of 
weakness, and the cause of this is not: far to seek. On exposing 
the joint it was found that while the fishplates perfectly fitted 
one rail, they did not, owing to a slight variation of section, fit 
the corresponding rail. How is this to be obviated? Only by 
milling both rail and fishplate, and it is only by this means that 
a perfectly mechanical fit can be obtained with fishplates. 
The experiment at Newtown was considered so satisfactory that 
it was determined to extend the principle, with further improve- 
ments, and in March last the work of relaying from Bridge- 
street to Hunter-street, a length of 15 chains, was commenced. 
The track to be removed consisted of 714 ibs T rails originally laid 
with 33 Ibs. separate guard, the guard on one side forming the 
fishplate. These guards, owing to their worn state, were removed 
four years ago, and old 42 ibs. rails substituted. The sleepers 
were spaced 3’ apart, and bedded in concrete. The rails had 
been down 10 years and 10 months, and when taken up and 
weighed had lost an average weight of over 20 ibs. per yard. 
The rails selected for relaying were 80 tbs. steel T, 30’ long 
laid on sleepers 2’ 4” apart, or thirteen to a rail, with broken 
metal ballast. For guards 42 Ibs. rails no longer suitable for 
