26 BULLETIN 1179, U. S. DEPARTMENT OF AGRICULTURE. 
An inspection of Table 11 shows that as a rule considerably better 
results were obtained where the crucibles were kept covered so that 
the oxidizing gases were not allowed to come in contact with the 
charge. In most instances where a molten slag was obtained and 
the heating continued sufficiently long, a very high percentage of the 
phosphoric acid contained in the charge was volatilized. When the 
crucibles were left open, however, it was noticed that a thick white 
crust or ''bloom" usually formed over this slag and unless the 
temperature was raised to the point where the fire-clay crucible itself 
fused, this crust gave no sign of melting into the slag beneath. The 
amount of this crust in a number of instances was fully as great as 
that of the underlying slag and showed upon chemical examination 
a higher percentage of phosphoric acid than that present in the 
original mixture. The ratio of lime to phosphoric acid in this unfused 
material conformed closely to the ratio of these ingredients in calcium 
pyrophosphate. Apparently under the oxidizing conditions existing 
at the surface of the mass the phosphoric acid distilling from the 
interior of the charge was fixed at the surface, producing calcium 
pyrophosphate according to the following equation: 
2Ca 3 (P0 4 ) 2 + P 2 5 = 3Ca 2 P 2 7 . 
It was found, however, that when the crucibles were kept covered 
until the fusion was well under way and the carbon or coke thus 
protected in the molten slag the covers of the crucibles could be 
removed without the formation of this crust, and the reaction 
continued to an end in spite of the oxidizing conditions at the surface 
of the slag. The addition of small quantities of alumina to the charge 
contributed somewhat to the fluidity of the slag, it being a well-known 
fact that the presence of this substance in small amounts lowers the 
melting point of both acid and basic slags. 
In these preliminary experiments the ratio of silica to lime was 
approximately 61 to 39, which is very close to the proportions of these 
ingredients in calcium trisilicate (Ca 2 Si 3 8 ). Charges having this 
silica-lime ratio gave slags from wmich a quicker and more complete 
evolution of P 2 5 was obtained than from those of a less acid nature. 
Such slags, however, congeal so readily that the tapping of the fur- 
nace is rendered somewhat difficult. Subsequent experiments indicate 
that the more nearly neutral slags are more readily fused and probably 
better adapted for actual furnace operation. 
These laboratory experiments pointed conclusively to the impor- 
tance of maintaining reducing conditions in the phosphate charge until 
fusion has taken place, and it appeared at first that the most feasible 
method of doing so in a mass containing much finely divided material, 
such as the pebble phosphates of Florida or the run-of-mine phosphates 
from Tennessee, was by the indirect heating of the charge so that 
the oxidizing gases from the burning fuel would not come in contact 
with the reacting mass until the latter was brought to a molten 
condition. Accordingly, a fire-brick furnace of semicommercial size, 
of the type shown in Figure 10 and described by the senior author 
and others in United States Patents Nos. 1241971 and 1282994, 
was constructed at Arlington Experimental Farm, Va. This furnace 
comprised a central or inner chamber (holding about 150 pounds of 
chirge) open both at the top and the bottom but constricted some- 
what at the lower end to prevent the charge from working through 
