The possibility that stars of the upper main sequence possess weak win
ds (of order 10(-14) to 10(-12) M-. yr(-1)) that may compete with atom
ic diffusion has been suggested by several authors, particularly by Ba
bel & Michaud (1991a) and Babel (1992). Such winds may affect the chem
ical abundances that appear in the atmospheres of these stars. However
, obtaining unambiguous evidence for the existence of a weak wind is n
ot easy. We consider the possibility that naturally abundant chemical
elements such as C, O and Ne, which are nor expected to be made overab
undant in a stellar atmosphere by radiative levitation, may accumulate
in or near the photosphere of a mass-losing star, thus serving as tra
cers of mass loss. We first show that the most abundant elements are n
ot expected to be overabundant in the atmospheres of main sequence sta
rs in the range 8000 less than or equal to T-e less than or equal to 1
5000 K due to radiative acceleration. We next show that hydrogen mass
loss from the atmosphere, at a rate in the range of roughly 10(-14) le
ss than or equal to (M) over dot less than or equal to 10(-12) M-. yr(
-1), could lead to accumulation of a few abundant elements in or not f
ar below the stellar photosphere, provided that any turbulence in the
wind at the level of the photosphere is not sufficient to prevent diff
usion. We find that such mass loss in late B stars should lead to obse
rvable overabundance of Ne, while in early A stars, such mass loss cau
ses O to accumulate in the atmosphere. Detection of overabundances of
these elements would provide direct evidence of the presence of weak m
ass loss. The few available Ne abundance determinations do not allow u
s to decide whether such weak winds are present in any B stars, but nu
merous measurements of O abundance imply that either winds in this mas
s loss range do not occur, or else that atmospheric diffusion is preve
nted by turbulence in early A stars.