Yohkoh X-ray observations during the several-hour decay of six large solar flares show that the main emitting region is in the form of a bright loop-top source whose spatial extent well after the impulsive stage explains most of the broadening of S XV and Ca XIX X-ray lines, which would otherwise imply nonthermal velocities of 100 km s-1 or more. During the decay of each event, the electron temperature and emission measure decline slowly. For the nearly disk center event of 1992 February 27, the volume increases and the deduced lower limit to the electron density Ne decreases from ~1011 cm-3 to just below 1010 cm-3. The nonuniform emission along the loops indicates that Spitzer heat conduction does not operate. If radiation is the only or main energy-loss mechanism, the lower limits to Ne imply upper limits to radiation cooling times of up to 10 hr, less than the flare decay time, so necessitating a continuing energy input. Although magnetic reconnection of previous broken open loops (the Kopp-Pneuman model) has been widely regarded as the source of energy in such flares, the small nonthermal velocities—only 20-50 km s-1 as indicated by SUMER observations—and the lack of spectroscopic indicators of large inflows or outflows argue against it. Other flare models must also explain the relatively small nonthermal velocities. An alternative picture of loop oscillations recently seen in TRACE and Yohkoh data is proposed for the production of observed nonthermal velocities.