BY A. H. S. H^CAS. 677 



I then made an exact determination of the amount of KCl. 

 The samples selected were free from sand. Jn one sample, the 

 percentage of KCl was 6055, and in a second it was 61 '24. 



Speaking of the analysis to my friend, Mr. T. Steel, who has 

 great laboratory experience in organic analyses, he evinced great 

 interest in the matter, and offered to make a complete and ex- 

 haustive analysis of the efflorescence. 



Mr. Steel's analysis gave: - 



KCl 58-20 



CaSO^ 

 Mannitol 

 Sand 

 Water 



trace 

 36-70 

 4-90 

 0-30 



100-10 



He estimated the potassium and the chlorine separately, and 

 thus definitely proved the complete absence of sodium. 



If we deduct the sand, the percentage of KCl in the efflor- 

 escence is 61 -3 and that of the matinitol 38-6. These percentages 

 agree well with the foi-mula of an exact chemical compound, 

 CfiHjj(OH)i;, 4 KCl, as was pointed out by my colleague, Mr. 

 Carpenter. 



I then tried if this compound could be formed artificially in 

 the laboratory. The chloride and mannitol were dissolved in 

 the above proportions in a small quantity of water and evaporated 

 down (1) rather quickly, and (2) very slowly in a desiccator. In 

 the former case, the two substances crystallised out separately 

 in their characteristically different crystalline forms. In the 

 second case, crystals of one form only were obtained, needles 

 ai-ranged in beautiful feathery gi-oups, crystals appai-ently 

 identical with the crystals of the efflorescence. 



I tried mixtures of alkaline chlorides with mannitol corres- 

 ponding to the formula CeHgfOH),,, 2KC1; C,;Hs(OH)fl, 4NaCl; 

 C,;H„(()H)ii, 2NaCl. In all cases, the substances crystallised 

 out independently. Ko compound was formed. 



The compound CRHg(OH)„, 4KC1 is unstable in the presence 

 of water, which dissolves more of the mannitol, and of alcohol, 



