In convection systems with boiling springs, geysers, fumaroles, and other thermal features, the modes of heat flow become increasingly complex as a single liquid phase at depth rises into the near-surface environment where heat flows by convection of liquid and vapor and by conduction in high thermal gradients. This paper is mainly concerned with the changing patterns of conductive heat flow as related to channels of subsurface convective flow and to horizontal distance from spring vents. The primary data consist of temperatures measured in 13 cored drill holes as drilling progressed. Some temperatures plot convincingly on straight-line segments that suggest conductive gradients in rocks of nearly constant thermal conductivity. Temperature gradients and the conductive component of total heat flow nearly always decrease drastically downward; the gradient and heat flow of the lowest depth interval recognized in each hole is commonly only about 10 percent of the highest interval; the changes in gradient at interval boundaries are commonly interpreted as channels of near-boiling water or of cooler meteoric water. Temperature reversals are probably related to inflowing cooler water rather than to transient effects from recent changes. Some temperatures plot on curved segments that probably indicate dispersed convective upflow and boiling of water in ground penetrated by the drill hole. Other similar curved segments are too low in temperature for local boiling and are probably on the margins of hot upflow zones, reflecting conductive cooling of flowing water. The conifers of Yellowstone National Park (mainly lodgepole pine) seem to have normal growth characteristics where near-surface conductive heat flow is below about 200 heat-flow units (1 HFU = 10^-6 cal/cm² = 41.8 mW/m²). Most areas of abnormal "stunted" trees (low ratio of height to base diameter, and low density of spacing) are characterized by conductive heat flows of about 250 to 350 HFU. The critical factor affecting growth is probably the seasonal maximum soil temperature at the root depths preferred by each form, rather than the heat flow as such. Heat flows up to 300 HFU are greatly dominated near the surface by conduction and are little affected by convection within the measured intervals. With increasing total heat flow above 300 HFU, the convection component, as indicated by snowfall calorimetry, becomes increasingly important. Snowfall calorimetry and tree growth as related to heat flow are calibrated by the heat-flow data considered here. Both snowfall calorimetry and tree growth patterns can be extrapolated rapidly, although without high precision, to thermal areas that lack subsurface data.