Data presented in this paper show that spring temperatures have a consistent relationship to air and ground temperatures within a restricted area. Spring temperatures tend to increase with decreasing elevations, just as air temperatures do. Ground temperatures obtained by extrapolating temperatures measured in drill holes in the vicinity of Crater Lake are similar to air temperatures, but the data obtained by Powell and others (1988) show that this relationship is not universal. The trend of the cold-spring data makes it possible to show that temperatures and specific conductances for the Wood River group of springs are anomalous.

The data for the springs in the Wood River group challenge a simple numerical temperature criterion for what is a thermal spring. Compared to other springs in the vicinity of Crater Lake, they are anomalous in temperature, dissolved chloride, sulfate, boron, and lithium, and chloride flux. The magnitude of the anomaly is small for all measures except for chloride flux; however, it is possible that there is a parent water that is high in all measures that is substantially diluted in the samples that have been obtained. The broad similarity of the characteristics of the springs of the Wood River group and the inflow to Crater Lake is intriguing, especially in view of the location of most of the springs of the Wood River group on a fault that trends towards Crater Lake but cannot be followed ounder the cover of the young volcanic rocks of Mount Mazama and the climactic eruption (Kienle and others, 1981, Sherrod and Pickthorn, in press). By having data on cold-spring temperatures at various elevations, it has been possible to establish that the springs of the Wood River group represent a thermal anomaly. The spring temperatures do not satisfy the numerical criterion for thermal springs, but they do satisfy Meinzer’s appreciable and Waring’s noticeable criteria because of the amount of data available to define background temperatures. For most purposes, the use of the numerical criterion is to be preferred, because background data usually are not available for the more careful comparison done for this area.