The Mercury Pool Interferometer

Albert Michelson, Edward Morley, and the Missing Luminiferous Ether

Resting inside an annular cast-iron trough bolted to a brick pier in the basement dormitory of Adelbert College in Cleveland, Ohio, a square slab of Berea sandstone five feet across and one foot thick floated on a half-inch pool of liquid mercury. Mounted at the four corners of the two-ton stone block stood sixteen polished speculum-metal mirrors arranged to bounce two split beams of yellow sodium flame light back and forth across the diagonals before recombining them in a small brass viewing telescope. Over six hours of July 8, 9, 11, and 12, 1887, the physicist Albert Abraham Michelson and the chemist Edward Williams Morley walked slowly around the stone pier in a circle, keeping their eyes glued to the moving eyepiece without ever touching the sandstone block as it glided smoothly through $360\text{ degrees}$ on its frictionless mercury film.

Throughout the nineteenth century, James Clerk Maxwell's equations of electromagnetism appeared to require an invisible, all-pervading mechanical medium—the luminiferous ether—through which light waves propagated at velocity $c$, just as sound waves require air. If the ether sat stationary in absolute space while the Earth orbited the Sun at $v = 30\text{ kilometers per second}$, then every terrestrial laboratory was swept by a continuous thirty-kilometer-per-second "ether wind." A beam of light launched upstream and downstream parallel to Earth's motion would take slightly longer to complete a round trip of distance $L$ than an identical beam launched perpendicular to the ether wind.

As Maxwell himself had pointed out in an 1879 letter to the Nautical Almanac office, the round-trip time difference was almost unimaginably small. On the upstream-downstream leg, the speed advantage gained moving with the wind ($c + v$) is nearly cancelled by the speed penalty fighting against the wind ($c - v$) on the return trip. The first-order ratio $v/c$ ($10^{-4}$) cancels out completely, leaving only a second-order time delay proportional to $v^2/c^2 \approx 10^{-8}$—one part in one hundred million. No mechanical cogwheel or rotating mirror could resolve a one-hundred-millionth change in flight time.

Michelson solved that sensitivity bottleneck by using the light wave's own crests and troughs as the ruler. In his interferometer, a half-silvered glass plate set at a $45^\circ$ angle split a single coherent beam of yellow sodium light (wavelength $\lambda = 589\text{ nanometers}$) into two perpendicular arms. When the two beams bounced back and overlapped at the detector, any difference in their round-trip travel time shifted their relative phase, moving a zebra-striped pattern of bright constructive and dark destructive interference fringes across the field of view. Rotating the entire table by $90^\circ$ swapped which arm faced into the ether wind, doubling the expected fringe shift $\Delta N$ across an effective optical arm length $L$:

When Michelson built his first brass-frame prototype in Hermann von Helmholtz's Berlin laboratory in 1881 and moved it to the brick vault beneath the astrophysical observatory at Potsdam, the instrument failed under environmental noise. With an arm length of only $1.2\text{ meters}$, the expected fringe shift was just $0.04\text{ fringes}$. Worse, footsteps on the pavement a hundred meters outside the Potsdam building shook the brass arms, and when Michelson rotated the metal frame on its mechanical pivot, uneven mechanical strain and the warmth of his own hands warped the arms by a fraction of a micron, shifting the fringes by more than the entire ether signal.

Teaming up with Morley in Cleveland in 1887, Michelson redesigned every mechanical element around vibration isolation and optical multiplication. Replacing the hollow brass arms with a solid two-ton block of Ohio Berea sandstone gave the optical bench massive thermal inertia and structural stiffness. Floating that stone block in a circular trough containing sixty kilograms of liquid mercury eliminated solid-on-solid bearing friction altogether: a single tap of a finger set the two-ton block rotating for minutes without mechanical jerk or structural bending, while the dense liquid mercury damped out high-frequency seismic vibrations from horse-drawn wagons on Euclid Avenue outside. Finally, by reflecting each light beam back and forth eight times between the corner mirrors, Michelson folded an eleven-meter optical arm length ($L = 11.0\text{ meters}$) onto a $1.5\text{ meter}$ stone square, magnifying the predicted ether shift tenfold to $\Delta N = 0.40\text{ fringes}$.

Even with white-light fringes centered to a hundredth of a wavelength and readings logged at sixteen compass points during continuous rotation at both noon and six o'clock in the evening, the zebra stripes refused to budge. Instead of the predicted shift of 0.4 interference fringes, Michelson and Morley measured less than 0.01—the most consequential null measurement in physics, which forced Hendrik Lorentz and Albert Einstein to discard absolute space.