36 GROWTH IN TREES. 
equalizing variations, which decreased in amplitude with the advance 
of the season, until the record was but little away from a level line 
until January 1921, when the instrument was dismounted to be re- 
placed by one of improved pattern. The total increase during 1920 
amounted to 12 mm., or about one-half that of the previous year. 
The reduction of the measurements given was in fair agreement with 
measurements of cores taken from the tree in February 1921. The 
wood formed at the beginning of the season is characterized by ex- 
tremely large vessels, which serve to denote the part of the layer formed 
earliest in the spring.’ 
Vines has described the annual rings of Frazinus excelsior as being 
2 to 3 mm. in thickness and as consisting of an internal spring zone 
of wide vessels with wood-parenchyma and rather thin-walled woody 
fibers, external to which, later in the season, thick-walled woody 
fibers form with scattered, smaller vessels surrounded by wood paren- 
chyma. The wood formed at the close of the season consists chiefly 
of wood parenchyma and small, very thick-walled vessels. He notes 
that young trees of this species in a damp soil may form a layer of 
wood 15 mm. in thickness in a season, which is practically equivalent 
to that of the Arizona ash under conditions of irrigation. 
The daily variation in 1919 was greater than that seen at any other 
time or in any other tree. Thus the daily shrinkage in March might 
be as much as 0.4 mm., with an over-compensating increase of 0.6 mm. 
by the following morning; in April these measurements were 1.1 mm. 
and 1.4 mm. In May the variations rose to 1.6 and 2.1mm. Cor- 
related variations are displayed by elongating seedlings of the ash 
during the same months. 
The materials of which the floating frames were constructed were 
tested to determine whether or not some of this apparent variation 
might be due to shortening and lengthening of the metal bars and rods 
under changing temperatures. 
Calibrations of a bar of bario, such as was used in this instrument, 
showed that its variation was 0.0000065 per unit part with 1°C. The 
length of the axis of the yoke was 14.5 inches or 363 mm., and its 
variation would therefore be 0.000025 mm. for 1°, and 0.0005 mm. 
for 20°, which would represent the possible total in any one day. This 
amplified 10 times by the levers would amount to about 1/200 of one 
space on the record sheet. This would be wholly compensated by the 
expansion and contraction of the short contact screws and by the varia- 
tion in the instrument, so that for the purposes of this research the 
instrumental error may be neglected. 
1 The above calculations are to be used in correction of the statements in the Carnegie Inst. 
Wash. Year Book for 1919, p. 74. 
2Vines, 8. H. Text Book of Botany, p. 198. 1895. 
