312 Ohservalions 07i /?ic occasional Appearance of Water 



cannot but suppose that these natural crystals, like those of 

 art, had at first a similar excess of water of crystallization, 

 which through a series of ages had escaped by capillary at- 

 traction through the minute pores of the mineral. 



It was whilst attending Professor Jameson's lectures on natu- 

 ral history, in 1810, that I fii'st directed my attention to these 

 curious specimens ; and having often heard various accounts 

 given of empty bottles well corked, which had been sunk in 

 the ocean and brought up full of water, for which I could only 

 account bv granting the porous nature of glass, I began to 

 suspect that all siliceous bodies might be porous. As the ex- 

 periments with empty bottles were in general not properly at- 

 tested, and were often contradictory, it became necessary to 

 have them repeated with care and minuteness : several friends 

 sailing for different parts of the East and West Indies, were 

 ready to promise me the requisite information, which was to 

 be tried upon thick glass globes about five or six inches dia- 

 meter hermetically sealed at the glass-house as soon as made, 

 from which circumstance each presented a partial vacuum ; 

 (see figure 1 ; abed are the notches by which it was attached;) 



fig. 2. 



fis. 3. 



but unfortunately the promised experiments were either mis- 

 managed, or forgotten in the hurry of commercial concerns. 



In 1813 my attention was recalled to the subject by the fol- 

 lowing appearance, which occurred during some experiments 

 with boiling water. Upon inadvertently pouring hot water 

 into a crystal decanter, which had a crack at the mouth about 

 three inches in length, I observed it dart downwards to nearly 

 five inches fi"om the top ; when I ceased pouring in the hot 

 water, the crack ascended again to the thiixl inch; the appli- 

 cation was repeated many times, and presented the alternate 

 extension and contraction of the crack each time as at first. (See 

 fig. 2; ef shows the crack, and the dotted line_/o- shov/s the 

 extended ci'ack which closed up again.) Upon applying pres- 

 sure to the top of pieces of window glass, laid flat upon a table, 

 I found small Qracks at the edges also alternately advance and 

 retire. (See figs. 3 and 4 ; h i is the original crack, ik the exten- 

 sion 



1 



