I5    Hafele and Keating Travel Around the World


Around 1960 the accuracy of cesium atomic clocks was so great that one could consider directly verifying the effects of STR and GTR with such clocks. While other experiments using satellites were being planned, J. Hafele and R. Keating quietly prepared a test using ordinary commercial airliners. In October 1971 they flew - first in an eastward direction and then in a westward direction - around the world. Together with the 4 atomic clocks they occupied 4 first-class seats. During the flight they continuously recorded their altitude, speed and direction.

GTR requires clocks at higher altitudes to run faster than identical clocks on the ground. STR requires that more time elapse for a clock at rest than for one moving. One must, however, be in an inertial frame which the rotating earth is not! The clocks on the ground have as a result of the earth’s rotation a significant speed (given that they are not at the North Pole ...). This speed must be added to the speed of the aircraft during the eastward flight. When flying westward it must, however, be subtracted, which after extensive analysis leads to the following table of expected time differences compared to the clocks ‘left behind’:

Predicted Effect Flying East Flying West
GTR (Gravitation) mmmm mmmm + 144 ± 14 ns  mmmm mmmm + 179 ± 18 ns mmmm
STR (Velocity) - 184 ± 18 ns + 96 ± 18 ns
Total - 40 ± 23 ns + 275 ± 21 ns

The +273 ± 7 ns measured by Hafele and Keating on the westward flight are in (almost suspiciously) good agreement with the expected value. On the eastward flight the clock with serial number 361 and its -74 ns seemed to be marching to its own drummer, but the other three with -51 to -57 ns were in harmony. The average of all four clocks was - 59 ± 10 ns.

Particularly nice aspects of this experiment are, firstly, that the effects of the STR and GTR can in a certain sense be separated, although a given clock is always measuring the total impact. Secondly, it is simply cool how these two with minimalistic resources stole the show from the others with their expensive satellite experiments.

Hafele and Keating could confirm the predictions of the GTR and STR with an accuracy of about 9%. A significant increase was achieved in the experiment of Maryland, which we discuss in I6. With a rocket flight with a hydrogen powered maser on board, which served as the clock, R. F. C. Vessot and M.W. Levine finally reached in 1979 after several years of data analysis (!!) a result with an uncertainty of ±0.02% which is within the limits of the predictions of the STR and GTR. This rocket is now flying under the title ‘Gravity Probe A’ (I8). Today these clock experiments are performed, so to speak, in the opposite direction: We assume that the atomic clocks in the GPS satellites are ‘wrong’ according to STR and GTR, correct the time signals accordingly, and generate the 3D position near the earth’s surface to within a few centimeters based on precise orbital data and the corrected signals from four GPS satellites (see I7).

Addendum May 2009: My skepticism concerning the too-good-to-believe match between prediction and experimental results seems to be confirmed. A.G. Kelly points out in an Internet post that the recently published original data of Hafele and Keating in no way allow a serious derivation of the results which made the two famous. Instead of a 'cool' or 'clever' experiment we should perhaps rather speak of fraud.



Hafele and Keating with their atomic clocks on board their Boeing 747 flight