260 Dr. Beddoes's Observations on the Process for 
EXPERIMENT III. 
A coated flint glass retort was employed in this instance. 
The apparatus resisted a strong heat for two hours ; and air, 
of the heavy inflammable kind, came constantly over. 
I 
EXPERIMENT IV, 
A coated retort of crown glass, containing six ounces 
Troy of the same iron, was placed on a crucible nearly full 
of sand, and disposed as in the former experiments. I now 
wished to measure the quantity of the air, and I therefore de- 
termined to receive it in mercury. It would have been in 
vain to attempt this in water, on account of the carbonic acid 
air. About twelve o'clock the retort was judged to be of a 
dull red heat, and inflammable air came over. The orifice of 
the transmitting glass tube was now covered to the depth of 
half an inch with mercury, when the discharge of air instantly 
ceased : the lute seemed entire. Some of the mercury being 
removed, so as to leave just, enough to cover the mouth of the 
tube, immediately the air issued again in bubbles, a proof 
that the apparatus was entire. The mercury was poured into 
the trough again, and in an instant there was a cessation of air. 
The. mouth of the tube being uncovered, and a lighted paper 
applied, a blue flame appeared, and continued to burn, so 
great was the quantity of air discharged. The orifice of the 
tube was one-tenth of an inch in diameter. We found that 
this constant flame could be produced at any time during three 
hours and' an half. When water was substituted in place of 
mercury, air issued slowly, and as if with difficulty, under a 
pressure of five inches. When only half an inch was left over 
the mouth of the tube, small bubbles ascended freely . During 
