Later, in 1887, Mr. T. K. Tapling, writing in _Le Timbre-Poste_,[42] claims the difference cannot be due to shrinkage of paper because the stamps have all shrunk evenly, and attributes it to some defect in the process of making the plates. He reasons thus:-- Les timbres sur les feuilles de n'importe quelle valeur étaient tous identiques comme type. Ils furent gravés sur acier, je pense par MM. Perkins Bacon et Co., chaque timbre par un procédé de réduplication, étant reproduit d'une matrice; la planche étant ensuite durcie pour l'impression. Il n'y a par conséquent pas de variété de types, les lignes des gravures sur les timbres courts étant les mêmes que celles sur les timbres longs, excepté qu'elles sont un tant soit peu contractées.... Il me semble plus que probable que la différence en longeur des exemplaires puisse être attribuée à un léger défaut dans le procédé de réduplication des planches de la matrice originale. [42] =Le Timbre-Poste=, Numéro Jubilaire, page XXXV. As a matter of fact the stamps did not shrink evenly, but very unevenly. Mr. W. B. Thornhill, writing on these same stamps in 1889,[43] says:--"You can hardly find two stamps of exactly the same measurements in the same value, though the difference in many cases is too small to signify"; and he proceeds to show the extreme variations in a carefully prepared table including every value on every variety of paper for issues from 1855 to 1867. The greatest variation in the vertical dimension seems to be about 1 mm. in 26 mm., or roughly 4%, and in the horizontal dimension about 1/4 to 1/2 mm. in 19 mm. or roughly 1-1/4 to 2-1/2%. These dimensional differences being so palpably existent, therefore, what factors are we to consider in looking for their cause? There seem to be but three: first, an original die or matrix for each different size; second, one original die only, whose impressions on the printing plate show variations resulting from the process of transferring them; third, a printing plate with all the impressions exact duplicates of the one original die, but whose reproductions in ink on dampened paper are varied by the shrinkage of the paper in drying. [43] =Philatelic Record=, XI: 71.