20-2 ASCENT OF KAZBEK. 



We were unanimously of opinion tliat an attempt to 

 return by our morning's route would end in disaster, and 

 that a way must be sought in another direction. This 

 could only be on the northern flank of the mountain, and it 

 was satisfactory to see that, for a long distance on that 

 side, there was no serioas difficulty. A steep slope of snow 

 (not ice) fell away from our feet to a great neve-plateau, 

 which we knew must pour down glaciers into the glens 

 which open into the Terek valley below the Kazbek station. 

 A very few minutes' consideration determined us to follow 

 this line, abandoning for the time our camp and the 

 porters on the south side of the mountain. The first haU'- 

 dred feet of descent down the hard snow-bank were steep 

 enough ; I was ahead, and neglected to cut good steps, an 

 error which resulted in Moore's barometer getting a jolt 

 which upset it for several hours. Hapx^ily, the little thing 

 recovered during the night, and told us our approximate 

 heights for many a day afterwards. Very soon the slope 

 became gentle enough to allow us to dispense with axe- 

 work, and we trudged straight and steadily downwards, 

 until we were almost on the level of the extensive snow- 

 fields upon which we had looked from above. Here we 

 again halted, to consider our further course. We were on 

 an unlaiown snow-plain, at a height of 14,000 feet above 

 the sea, and it was most undesirable to hazard our chance 

 of reaching terra cognita ere nightfall by any rash or 

 hasty move. One plan suggested was to turn to the left, 

 and cross a pass we had good reason to believe connected 

 the plateau we were on with the neve of the glacier by which 

 we had ascended. This course, if successfully carried 

 out, would have brought us back to our tent and baggage, 

 but its probable length was a fatal objection. Eventually 

 we determined to keep nearly due north, across the snow- 

 field, towards a ridge which divides two glaciers flowing 

 into different branches of the glen of Devdorak. We des- 



