We present the first unambiguous case of external variability of a radio gravitational lens , CLASS B1600+434 .
The Very Large Array ( VLA ) 8.5–GHz difference light curve of the lensed images , taking the proper time-delay into account , shows the presence of external variability with 14.6– \sigma confidence .

We investigate two plausible causes of this external variability : scattering by the ionized component of the Galactic interstellar medium and microlensing by massive compact objects in the bulge/disk and halo of the lens galaxy .
Based on the tight relation between the modulation-index ( fractional rms variability ) and variability time scale and the quantitative difference between the light curves of both lensed images , we conclude that the observed short-term variability characteristics of the lensed images are incompatible with scintillation in our Galaxy .
This conclusion is strongly supported by multi–frequency Westerbork Synthesis Radio Telescope ( WSRT ) observations at 1.4 and 5 GHz , which are in disagreement with predictions based on the scintillation hypothesis .
Several arguments against scintillation might need to be reevaluated if evidence is found for significant scatter-broadening of lensed image B seen through the lens galaxy .
However , the frequency-dependence and time scale of variability from image A are not affected by this and remain strong arguments against scintillation .

On the other hand , a single superluminal jet-component in the source , having an apparent velocity 9 \la ( v _ { app } / c ) \la 26 , a radius of 2–5 \mu as and containing 5–11 % of the observed 8.5–GHz source flux density , can reproduce the observed modulation-indices and variability time scale at 8.5 GHz , when it is microlensed by compact objects in the lens galaxy .
It also reproduces the frequency-dependence of the modulation-indices , determined from the independent WSRT 1.4 and 5–GHz observations .
The difference between the modulation-indices of the lensed images ( i.e .
2.8 % and 1.6 % at 8.5 GHz in 1998 for images A and B , respectively ) , if not affected by scatter-broadening of image B by the ionized ISM of the lens galaxy , can be explained through a different mass-function for the compact objects in the bulge/disk and halo of the lens galaxy .
Comparing the observations with microlensing simulations , we place a tentative lower limit of \ga 0.5 M _ { \odot } on the average mass of compact objects in the halo line-of-sight .
The above-mentioned set of mass-function and source parameters is consistent , although not unique , and should only be regarded as indicative .

The only conclusion fully consistent with the data gathered thus far is that we have indeed detected radio microlensing .
The far reaching consequence of this statement is that a significant fraction of the mass in the dark–matter halo at \sim 6 kpc ( h =0.65 ) above the lens–galaxy disk in B1600+434 consists of massive compact objects .