We present near-infrared observations of the counterpart of GRB 000301c .
The K ’ filter ( 2.1 \ > \mu m ) light curve shows a well-sampled break in the decay slope at t \approx 3.5 days post-burst .
The early time slope is very shallow ( \sim - 0.1 ) , while the late time slope is steep ( -2.2 ) .
Comparison with the optical ( R band ) light curve shows marginally significant differences , especially in the early time decay slope ( which is steeper in the optical ) and the break time ( which occurs later in the optical ) .
This is contrary to the general expectation that light curve breaks should either be achromatic ( e.g. , for breaks due to collimation effects ) or should occur later at longer wavelengths ( for most other breaks ) .
The observed color variations might be intrinsic to the afterglow , or might indicate systematic errors of \gtrsim 0.08 magnitude in all fluxes .
Even if the break is achromatic , we argue that its sharpness poses difficulties for explanations that depend on collimated ejecta .
The R light curve shows further signs of fairly rapid variability ( a bump , steep drop , and plateau ) that are not apparent in the K ’ light curve .
In addition , by combining the IR-optical-UV data with millimeter and radio fluxes , we are able to constrain the locations of the self-absorption break and cooling break and to infer the location of the spectral peak at t = 3 days , f _ { \nu } \approx 3.4 \ > \hbox { mJy } at \nu \approx 10 ^ { 12 } \ > \hbox { Hz } .
Using the multiwavelength spectral energy distribution , we are able to constrain the blast wave energy , which was E \gtrsim 3 \times 10 ^ { 53 } \ > \hbox { erg } if the explosion was isotropic .
This implies a maximum gamma ray production efficiency of \sim 0.15 for GRB 000301C .