Recent Ulysses observations of millisecond spikes superposed on broader Langmuir wave packets in type III radio sources are compared quantitatively with constraints from the theory of wave collapse. It is found that both the millisecond spikes and the wave packets have fields at least 10 times too small to be consistent with collapse, contrary to previous interpretations in terms of this process. Several alternative explanations are considered and it is argued that the spikes should be interpreted as either non-collapse phenomena or observational artifacts. To the extent the observations are representative, this rules out theories for type III bursts at ~1--4 AU that rely on collapse. ¿ American Geophysical Union 1995