p. Groom 15 



the ends of the board, because the water was entering mainly through 

 these. 



Second, the boards actually shortened while at the same time they 

 continued to widen. 



Third, once more lengthening took place and was accompanied by 

 widening. 



During drying the reverse phenomena wholly or in part were 

 exhibited; for the shortening at the beginning was succeeded by 

 elongation, which was final in one case (yang 2) but was followed 

 by lengthening in the case of yang 1 ; in all these cases the board 

 continued throughout to shrink in width (excepting for occasional slight 

 deviations). 



Yet throughout these contrary changes in length the board 

 continued to increase in surface during the absorption of water, and 

 to decrease in surface during drying. And, in connexion with the 

 explanation of the phenomenon, it is worthy of note that the board 

 (tangential, yang 2) which showed the smallest ultimate decrease in 

 length and most persistent elongation during drying yet experienced 

 the greatest decrease in surface. 



The most probable explanation of the reversal in change of length 

 during the absorption of water appears to be one akin to that offered 

 by Professor Brereton Baker, who suggested the analogy of a net-work 

 or lattice-work in which swelling takes place more rapidly along the 

 length than at right angles to this. In this case the meshes formed 

 by the crossing vessels, fibres, and so forth are lozenge-shaped, narrow, 

 and elongated along the axis of the tree-trunk (and board). 



If we imagine two fibres or tracheae ABC and DBE intersecting 

 at B, and consider the two sides BC and BE of the triangle CBE, then 

 as water is absorbed by the board it is taken in most rapidly at the 

 ends and travels more rapidly (in the vessels) along ABC and DBE 

 than in any other direction. The result is relatively considerable 

 elongation of BC and BE, with relatively inconsiderable increase in 

 the angle CBE. Increased distribution of the water at right angles 

 to BC and BE next causes more rapid expansion in those directions, 

 tending to cause the two sides to rotate outwards about the centre B 

 and thus cause a widening of the base of the triangle but a shortening 

 of the perpendicular from B to that base. This would also take place 

 if the sides elongated to a disproportionately small extent. Thereafter 

 elongation of the two sides associated with disproportionately less 

 expansion at right angles would cause a reversal of the immediately 



