High precision VLA flux density measurements for the lensed images of 0218+357 yield a time delay of 10.1 ^ { +1.5 } _ { -1.6 } days ( 95 % confidence ) .
This is consistent with independent measurements carried out at the same epoch ( [ Biggs et al .
1999 ] ) , lending confidence in the robustness of the time delay measurement .
However , since both measurements make use of the same features in the light curves , it is possible that the effects of unmodelled processes , such as scintillation or microlensing , are biasing both time delay measurements in the same way .
Our time delay estimates result in a confidence intervals that are somewhat larger than those of Biggs et al. , probably because we adopt a more general model of the source variability , allowing for constant and variable components .
When considered in relation to the lens mass model of Biggs et al. , our best-fit time delay implies a Hubble constant of H _ { o } = 71 ^ { +17 } _ { -23 } km s ^ { -1 } Mpc ^ { -1 } for \Omega _ { o } = 1 and \lambda _ { o } = 0 ( 95 % confidence ; filled beam ) .
This confidence interval for H _ { o } does not reflect systematic error , which may be substantial , due to uncertainty in the position of the lens galaxy .
We also measure the flux ratio of the variable components of 0218+357 , a measurement of a small region that should more closely represent the true lens magnification ratio .
We find ratios of 3.2 ^ { +0.3 } _ { -0.4 } ( 95 % confidence ; 8 GHz ) and 4.3 ^ { +0.5 } _ { -0.8 } ( 15 GHz ) .
Unlike the reported flux ratios on scales of 0.1 ^ { \prime \prime } , these ratios are not strongly significantly different .
We investigate the significance of apparent differences in the variability properties of the two images of the background active galactic nucleus .
We conclude that the differences are not significant , and that time series much longer than our 100-day time series will be required to investigate propagation effects in this way .