118 THE FUTURE OF ARID LANDS 
finally their discontinuance. Individual investigations such as 
the study of Polacca Wash (8) have not been followed up. 
First it seems necessary to improve our understanding of the 
hydrologic relationships between intensity and amount of pre- 
cipitation, infiltration, and surface runoff, for combinations of 
soils and vegetation in semi-arid areas. The small experimental 
watersheds maintained by the Soil Conservation Service and 
Forest Service in New Mexico and Arizona are a step in this 
direction, but lack of adequate funds keeps this effort pitifully 
small relative to the need for such information. 
A second field of needed research is in the hydraulics of flow of 
sediment-laden water. Particularly deficient is our understanding 
of the nature of bed and bank roughness and the manner in which 
sediment in transport affects hydraulic resistance. In most 
ephemeral channels bed roughness is determined primarily by 
the dunes or ripples formed by moving sediment. We have few 
observations and no theoretical concepts on which to build an 
understanding of this phenomenon. 
A third broad field is in the mechanics of gully formation, 
including hydraulic forces, phenomena in the realm of soil me- 
chanics, and physiographic principles. 
Our own recent work has been concerned with these problems, 
and at least indicates some of the possible approaches which 
appear fruitful. The work began as a study of interrelations of 
discharge, width, depth, velocity, slope, and sediment in natural 
channels. On some of these parameters a plethora of data exists in 
the records of the regular stream-gaging stations. But in the 
existing network of measuring stations few measurements have 
been made on water and sediment flows in ephemeral streams 
draining I to 10 square miles. To obtain measurements for 
analysis, during three summers of work in Wyoming and New 
Mexico we chased thunderstorms, trying to reach a storm center 
in time to observe arroyos in flood. When flow was found we waded 
out into the arroyo and measured the depth, velocity, and width, 
and sampled the sediment load. Successive sets of measurements 
were made during the falling stage of the flow. Later, measure- 
ments of channel slope and bed material were made. 
