Accurate frequency comparisons at the 1/spl times/10/sup -15/ level

Increased fractional frequency uncertainties in comparing two cesium-fountain primary frequency standards due to a long-baseline comparison process are examined. These include frequency uncertainties introduced by the time-transfer process and uncertainties introduced by possible long-term dead time in the fountain operation. Using common-view GPS it may take up to 40 days to reduce the time-transfer fractional frequency uncertainty to 1/spl times/10/sup -15/ A combination of common-view GPS and two-way satellite time and frequency transfer can reduce this to 24 days. With a very low noise local oscillator such as a cavity-tuned hydrogen maser, distributed fountain dead times as large as 87% can be present and yet contribute only an additional uncertainty of less than 3.3/spl times/10/sup -16/.