Assessment of random and systematic errors in microwave and submillimetre dielectric measurements

By considering the variation between runs comprising a given experiment, between the average results of independent experiments and between the extrapolated results of different experimental methods, some assessment of the systematic and random errors to be expected in high-frequency dielectric measurements is obtained. The liquid investigated is chlorobenzene as this has good chemical properties, possesses a rigid molecule and has been extensively investigated over a wide frequency range by other workers. Our results show that the random errors are frequency dependent but have average values of ±1.5% for ¿ and ±0.1% for n but that the systematic errors may reach as much as ± 3% for ¿ and ± 0.6% for n in the difficult regions. By combining data over a wide frequency range one may arrive at continuous curves whose estimated errors are ±1% for ¿ and ±0.2% for n.