18 
BULLETIN 376, U. S. DEPARTMENT OF AGRICULTURE. 
wax. Both, the couplings and attendant follow nuts were recessed 
at an angle of 45° in order to effectively bind rubber gaskets. Joints 
between ends of iron pipe' units were made in the same way. A 
cement made of equal parts of beeswax and turpentine softened the 
rubber gaskets and gummed the pipe threads so that the joints were 
mercury-tight. This was doubly assured by winding sewing thread 
into the wax in the pipe threads. 
The inside of all metal pipes and connections was japanned three 
coats thick in order to prevent amalgamation with the mercury. 
Color injector. — The only practicable method of measuring the 
velocity of water in some of the pipes tested was by timing the passage 
of some color or chemical. After various tests fluorescein appeared 
to offer the best results. In order to inject the color into the pipe at 
the upstream gauge the " fluorescein gun" 
was developed (fig. 3). This is connected 
to the nipple C through the T connection 
D. At the downstream manometer the 
nipple C connects directly to the cock E. 
Mercury container. — The usual lead tor- 
pedo weights for the current meter were 
dispensed with and a combination mer- 
cury bottle and meter weight was con- 
structed (fig. 3). The details are evident in 
the illustration with the possible excep- 
tion of the surge walls which divided the 
bottle into small compartments so that the 
mercury gave no trouble by surging when the bottle was used as a 
meter weight. Likewise the small holes in the walls at the top of 
the weight offered small chance of losing all the mercury in case of 
accident. 
FIELD METHODS. 
Mercury, 
Datum-'* I _ 
Fig. 2.— Hydraulic principles of mer- 
cury manometer of U-tube type. 
CHOOSING A REACH TO BE TESTED. 
In order to be considered adaptable for field tests, a pipe must be 
practically water-tight (or the leaks measured) and of such length, 
without bends or obstructions, that the effect of errors is minimized — 
the longer the better. Gentle curves, both vertical and horizontal, 
were thought desirable, as their effect must be considered in the 
design of practically all wood-stave pipes. No distinct bend in any 
pipe was included in the reaches tested. Such a bend would cause 
an appreciable loss of head aside from friction loss. Some method 
of determining the mean velocity in the pipe must be available. 
