The most important part of this treatise is the account of the comparative structure of roots, to which we will return later, when discussing Grew's anatomical conceptions. With regard to the position of the plant in the soil, he held somewhat mystical views. He believed that the "air-vessels" or tracheal elements, tended to draw the plant upwards, and the roots to pull it downwards. He says, for example, that the upper part of the roots of most seedlings ascend, because the first leaves being large and standing in the open air, "the _Air-vessels_ in them have a dominion over the young _Root_, and so yielding themselves to the sollicitation of the _Air upwards_, draw the _Root_ in part after them." In 1675 appeared Grew's third botanical work, _The Comparative Anatomy of Trunks_, which dealt with stem structure, as the previous work dealt with root structure. There is, in the British Museum, a particularly interesting copy of this book, which is elaborately annotated in manuscript. From internal evidence it seems almost certain that this is the author's copy, corrected in his own handwriting[10]. Some, though not by any means all, of the corrections are identical with the alterations found in the 1682 edition. Above the first plate is written "vide ye Book Interleavd," and we may perhaps hazard the guess that in this copy we have Grew's first suggestions, whilst those which he finally adopted in the second edition were inserted in the interleaved copy whose whereabouts, if it still exists, is unknown at the present time. Pl. 6 shews a typical page from the annotated copy. At the foot we find the note "Air-Vessels out of Parenchyma, transformed, as Caterpillars to Flys," shewing that Grew had arrived at some idea of the formation of vessels. The whole section of the book to which this page belongs is very much remodelled in the 1682 edition, but the analogy just quoted is introduced and Grew proceeds accurately to describe the origin of vessels. "And as the _Pith_ it self, by the Rupture and Shrinking up of several _Rows_ of _Bladders_, doth oftentimes become Tubulary: So is it also probable, that in the other _Parenchymous Parts_, one single _Row_ or _File_ of _Bladders_ evenly and perpendicularly piled; may sometimes, by the shrinking up of their Horizontal _Fibres_, all regularly breakone [sic] into another and so make one _continued Cavity_."