A New Determination of the Free-Space Velocity of Electromagnetic Waves

Author(s): K. D. Froome
Source: Proceedings of the Royal Society of London. Series A, Mathematical and Physical
Sciences, Vol. 247, No. 1248 (Sep. 9, 1958), pp. 109-122
Published by: The Royal Society
Stable URL: http://www.jstor.org/stable/100591 .
Accessed: 05/08/2013 06:00

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].

The Royal Society is collaborating with JSTOR to digitize, preserve and extend access to Proceedings of the
Royal Society of London. Series A, Mathematical and Physical Sciences.

http://www.jstor.org

This content downloaded from 128.141.48.10 on Mon, 5 Aug 2013 06:00:09 AM

All use subject to JSTOR Terms and Conditions
 A new determination of the free-space velocity
of electromagnetic waves
BY K. D. FROOME, PH.D.
Standard8 Divi8ion, National Phy8ical Laboratory

(Communicated by G. B. B. M. Sutherland, F.R.S.-Received 25 March 1958)

[Plate 1I

The paper describes the determination of the free-space phase velocity of electromagnetic
waves in vacuo by means of a millimetre-wave interferometer capable of high precision. As
already provisionally announced (Froonze I958) the result is:
co= 299 792*50 + 0 10 km/s,
where the variation represents the standard deviation of a single determination in statistical
combination with estimated systematic errors.

1. INTRODUCTION

An advanced form of the 'four horn' millimetre-wave interferometer previously

described (Froome 1954) has been used to measure the phase velocity of electro-
magnetic waves with the highest accuracy possible by this method. The operating
frequency of the instrument was 72 006 Gc/s, corresponding to a wavelength of
approximately 4 mm.
As before, the basis of the determination consisted of the simultaneous measure-
ment of the free-space wavelength, and frequency, generated by a microwave
source. The interferometer was used for the wavelength measurement and was of
the 'two-beam' type. It employed a movable carriage supporting a pair of re-
ceiving apertures at opposite extremities, symmetrically disposed between an
identical pair of transmitting horns energized with microwave power conveyed
by waveguide from the common source. The axes of the four horns were carefully
alined and the motion of the carriage was also arranged parallel to this direction.
The two received signals were mixed together to produce interference, the result-
ant output undergoing a minimum for everyhalf-wave displacement of the carriage.
The separation between each transmitter and the corresponding receiving horn
was of the order of 10 m and as the effective aperture dimensions were squares of
about 6 cm side, the receiving horns were well within the Fraunhofer diffraction
field of the transmitters. The received wavefronts were thus approximately spheri-
cal, and of radius equal to the appropriate transmitting-receiving horn separa-
tion. It is the change of these radii with the position of the movable receiving
carriage that gives rise to the diffraction correction, the evaluation of which was
the primary function of the 24 Gc/s prototype equipment.
The principal innovation with the 72 Gc/s interferometer was the -use of a
cavity resonator refractometer for the direct determination of the refractive
index of the air in the neighbourhood of the interferometer. This replaced the
less accurate method of using separate observations of atmospheric pressure,
[ 109 ]

This content downloaded from 128.141.48.10 on Mon, 5 Aug 2013 06:00:09 AM

All use subject to JSTOR Terms and Conditions
110 K. D. Froome
temperature and humidity with refractive index formulae such as those derived
by Essen & Froome (I95I) or Essen (I953).
During the course of the new determination a small error was discovered in the
results obtained with the prototype apparatus. This error was due to the assign-
ment of too high a value to the 1 m length-standard used in it.
Consequently the published values of co obtained with the prototype are too
high by O02 km/s. All the conclusions arrived at concerning the operation of the
prototype instrument are, of course, unaffected by the change.
The movable carriage of the new interferometer was displaced through 970
half-waves (roughly 2 m, twice the prototype displacement and six times the order
of interference) by means of end-standards known in terms of a light-wave standard
of length. The value of microwave wavelength so obtained, when multiplied by
the air refractive index and the microwave frequency gave a vacuum phase velo-
city which had still to be corrected for the effect of diffraction before the true free-
space value could be derived. This was done by making seven velocity measure-
ments, each for a different transmitting-receiving horn separation.
By virtue of (a) the greater carriage displacement, (b) the shorter wavelength,
and (c) the improved location, all major error-producing influences arising from
the method of microwave transmission, propagation, an-d reception were reduced
relative to their effect on the prototype machine by more than one order of magni-
tude. It is interesting to note that the greatest single uncertainty in the whole
measurement arises from the use of the length-standards.

2. DESCRIPTION OF APPARATUS

Figure 1 shows in diagrammatic form all the essential details of the interfero-
meter and associated equipment. The source of microwaves was the first harmonic
of a Pound stabilized Q-band klystron oscillator operating at 36-003 Ge/s. The
greater part of the output (about 10 mW) from this oscillator was fed by means of
a waveguide switch into one of two silicon crystal distorter units tuned for maxi-
mum harmonic output at 72-006 Ge/s. One harmonic generator was used to supply
the interferometer itself, the other for operating the refractometer by means of
which the refractive index of the air in the neighbourhood of the equipment could
be measured. This refractometer has been described elsewhere (Froome I955).
A smaller portion of the klystron power was used for driving the Pound frequency
stabilizer and a minute fraction for frequency measurement.
The measurement of microwave frequency was accomplished by comparing the
klystron output against a high harmonic of a 5 Mo/s quartz crystal standard. The
5 Mc/s was multiplied in stages of two to five times up to 600 Mc/s and then fed
into a silicon crystal harmonic generator mounted in waveguide, so that the
harmonic at exactly 36 Gc/s could be mixed with a small fraction of the klystron
output. The beat frequency between the two was detected by means of a cali-
brated communications receiver. When the refractometer was being used the
klystron frequency had to be variable (instead of constant at 36&003Gc/s) between
36-002 and 36-014 Gc/s and was set to the nearest 100 ke/s calibration point on the
communications receiver corresponding to approximate resonance of the refracto-

This content downloaded from 128.141.48.10 on Mon, 5 Aug 2013 06:00:09 AM

All use subject to JSTOR Terms and Conditions
 Free-space velocity of electromagneticwaves 111
meter cavity, the setting to perfect resonance then being made by means of a
small tuning plunger attached to the cavity, in the manner already described
(Froome 1955; Essen I953; Essen & Froome I950). The accuracy of frequency
determination was always at least as good as 1 part in 108 and was thus the most
precise measurement associated with the whole velocity determination.
When the interferometer was being used to make a wavelength measurement
the 72-006 Ge/s output from the appropriate harmonic generator was guided to a
hybrid junction ('magic-T'), which served as beam divider, and eventually
through two long arms of cylindrical waveguide to the transmitting horns. The
matching-stub to the left of the beam divider, together with the constant phase

freuecye6vis frqunc
horn~ ~~~r\

hybrid iuinction
(beam davider)
r ~~~~~~~wave
tra'nsmitting 1< uidel
ma7m -I

_ _
__ _ _ _ _ _ movable carriage 6

FIGURE 1. Diagram of 72 Ge/s interferometer.

waveguide interferometer (c.p.i.), constituted a device for altering the amplitude

of the energy transmitted down this arm without producing a phase displacement
(Froome I954). The phase-shifter to the right, together with the variable attenu-
ator, was required in order to adjust the position and balance of the first inter-
ference minimum on the movable receiving carriage.
After passing through the attenuators, the energy in each arm was transmitted
via a transducer (for the purpose of exciting the low-loss Ho, circular mode) to
cylindrical waveguide (12-5 mm bore) for transmission over the relatively long
journey to the transmitting horns. Before the energy entered each horn another
transducer served to convert to the H0omode of propagation in the standard-size
rectangular waveguide leading to the horn throat.
The movable part of the interferometer consisted of a pair of receiving horns
mounted on a large carriage situated centrally between the transmitting horns.
Both transmitting and receiving horns were identical in every respect and each
was fitted with a polystyrene lens at its mouth to correct the radiated waveform
(or equiphase receiving surface) to planar. The horn apertures were rectangles of
approximately 8 x 6 cm, the ]arger dimension being in the plane of the magnetic

This content downloaded from 128.141.48.10 on Mon, 5 Aug 2013 06:00:09 AM

All use subject to JSTOR Terms and Conditions
 112 K. D. Froome

field component, thereby making the radiation pattern as symmetrical as

possible.
The central platform of the receiving carriage held three wheels arranged to
run on a cast-iron bed of 3 m length taken from a precision measuring machine.
Beneath the surface of the bed ran a counterweight equal in mass to the carriage,
but arranged to move in the opposite direction so as to maintain a constant moment
on the concrete block supporting the bed and on the floor of the room beneath the
block (figure 2).

wire to pulley 72 Gc/s super-

and winch for heterodyne receiver
lifting end-guages attached to side
\,, of carriage 'honeycomb' wall

wooden screens
to receiving horns
receiving horn _ receiving horn
beam
carriage (gauge out)
2twolmn.en gaugesinlaggedboxgEx~

steel bed supporting E gauge

movable carrIage mounted on 'honeycomb wall
movable carriage lX

'honeycomb' wall across

counterbalance centre of tunnel
concrete block,
z4 h- floor
h * 8 X*S.S^^
^ >s-;~. \

8~~~~~~~~~~~~~~~~~~~~~~~
ft.it'hig

FIGURE 2. Method of mounting the receiving horn carriage of the 72 Gc/s microwave
interferometer in the duplex wind tunnel room.

The whole apparatus was installed in a very large wind-tunnel room. Across
the approximate centre of the room is a 'honeycomb' anti-turbulence wall and this
was used to form a convenient datum point for measurements of displacement of
the receiving carriage. Part of the carriage actually passed through a hole cut in
this wall. Figure 3, plate 1, is a photograph of the receiving carriage showing
the 'honeycomb' wall.
A highly sensitive electrical measuring head (T.T. & H. Electrolimit, divided
in units of 10p,in. (025fL) was mounted upon the 'honeycomb' wall in such a
manner as to be entirely independent of the steel bed (and concrete block) sup-
porting the carriage, except when a small ball mounted on the centre of the car-
riage (in line with the receiving horn axes) was in contact with the sensing anvil
of the Electrolimit head. Provision was made for an independent slow-motion
adjustment of the carriage.

This content downloaded from 128.141.48.10 on Mon, 5 Aug 2013 06:00:09 AM

All use subject to JSTOR Terms and Conditions
Froome Proc. Roy. Soc. A, volume 247, plate 1

*~~~~~~ M R |PI E

mi r C_

EU * . jl ._

*Mq 'WEb"ba
EUEEEIM
M*Ehu Ear 'C)

5 E~~~~~~

mum~~~~~~~~~~~~~~~~~~~(aip 112)
'.4

(Facing P. I 1 2)

This content downloaded from 128.141.48.10 on Mon, 5 Aug 2013 06:00:09 AM

All use subject to JSTOR Terms and Conditions
 Free-space velocity of electromagneticwaves 113
The energy picked up by the receiving horns was transmitted along oversize rect-
angular waveguide to a hybrid junction mixer mounted on the central platform
of the carriage. The 'mixed', or interference signal, was detected by means of a
superheterodyne receiver involving a second stabilized Q-band klystron and crys-
til harmonic generator as 'local oscillator'. This was affixed to the side of the
movable carriage (see figures 2, 3). The intermediate frequency, chosen to be
60 Mc/s, was first amplified by a wide-band 'pre-amplifier' and then by a v.h.f.
communications receiver with a final bandwidth of 5 kc/s. The output from the
detector stage of this receiver was indicated by a micro-ammeter placed close to
the receiving carriage slow-motion adjustment and the precision of setting on a
minimum was about 051-k.When the interferometer was not in use, the output
from the second Q-band klystron could be conveyed by means of a waveguide
switch to the detector associated with the refractometer.
The receiving carriage was constructed from aluminium tubing and it was found
that the positions of the receiving horns did not shift relative to the mixer by more
than + 5 [kin. as the carriage was traversed along the supporting bed. By use of
the Ho, mode of propagation in the oversize waveguide between receiving horns
and mixer (so that the guide phase velocity approaches the free-space value) the
effect of carriage distortion was further reduced because, providing the waveguide
follows the carriage distortion, the position of an interference minimum would be
largely independent of the positions of the receiving horns on the carriage.
The geometrical axes of the transmitting horns, the receiving horns, the ball
on the carriage, and the anvil of the Electrolimit gauge-head could be alined
accurately by means of an alinement telescope. The mounting of the receiving
horns was such that they could be rotated by 180? about their (horizontal) geo-
metrical axes; such rotation was part of the experimental procedure for eliminating
errors arising from any difference between the electrical and geometrical axes if
the carriage should happen to tilt a little as it is displaced. Corresponding errors
in the transmitting horns are of the same form as the leading term in the dif-
fraction cor. ection (see equation (1), ? 6) and were eliminated by the evaluation of
that correct .n.
The 2 m k ngth-standard used for measuring the carriage displacement through
970 minima (i.e. 970 half-waves) consisted of two 1 m end-bars set up in line, with
slip gauges wrung to each face and separated in the centre by a 6 mm steel ball.
When the end-standard was 'inserted' into position (i.e. the interferometer was
set on the 971st minimum) wheels attached at the Airy points of each gauge
automatically and accurately located it on rails fixed to the 3 m bed so that the
standard was free to move longitudinally as the setting on the 971st minimum was
made. The diameter of the 6 mm ball was measured by optical interferometry
under the load (200 g) imposed by the Electrolimit head. All slip gauges used were
likewise measured by optical interferometry.
The determination of the length of the two 1 m end-bars needs some elaboration.
This was accomplished by means of the apparatus of Sears & Barrell (I932), origi-
nally designed for the determination of the metre and the yard in terms of the
wavelength of the red line of the cadmium spectrum. With it one can measure a

8 Vol. 247. A.

This content downloaded from 128.141.48.10 on Mon, 5 Aug 2013 06:00:09 AM

All use subject to JSTOR Terms and Conditions
114 K. ID.Froome
special 1 m end-bar of X-section in terms of any suitable light-wave standard.
This X-bar can then be compared against other 1 m end-standards either by inter-
ferometry or by mechanical comparison. For the reasons given in ?8 the X-bar
was measured in terms of the cadmium red radiation as emitted by the interna-
tionally specified form of Michelson lamp.
When used on the microwave interferometer, the two 1 m end-bars were placed
inside well-lagged wooden boxes and fitted with a thermostatic control operated
by a sensitive contact thermometer controlling at 20 'C. The actual temperature
of the gauges was obtained by means ofi ten thermocouples related to a mercury-
in-glass thermometer calibrated to + 0-002 'C. The whole length-bar assembly could
be lowered into position by means of a winch and pulley arrangement.

3. THE CONSTANT PHASE WAVEGUIDE INTERFEROMETER (C.P.I.)

Although a full description of this device has already been published (Froome
1954), for the purpose of clarity a further summary of its mode of operation is
desirable. The basic theory is very simple: if two equal vectors, initially parallel
(equal waves in phase), are rotated in opposite directions through the same angle,
the direction of their resultant is unchanged while its amplitude is reduced.
movable slide

spring / ~~~~~~~~~~~~~~~mi

adjustment to ~ ~ ~ ~ ~ ~ I~ E-plane
~ ~ ~~~~~~ha
a rpilsto?nr l lossy sheet X

Fdjustsnent toUa E 4. Ai 'ta . s (c,p.i.)

depth of penetradao a c.p.i. t c i om
of lossvysheet b i /1 b rs

utlziga iun fjunction precision tubin ao

waegud
locking screws

lH-plane

output
FIGURE 4. Auxiliary constant phase inlterferometer (c.p.i.)

Figure 4 is a diagram of the c.p.i. It consists of a waveguide interferomneter

utilizing a hybrid junction for the dualpurposeof beam division and recombination.
Energy from the beam divider of the four-horn interferometer enters the E-plane
arm of the c.p.i. and at the centre of the junction divides into two equal parts and
enters the arms containing the mechanically linked movable-piston reflectors.
The waves reflected from these pistons interfere at the centre of the junction and
normally some energy is transmitted through the H-plane output arm and the
remainder is returned through the input arm. In figure 1 the matching stub
between c.p.i. and the main interferometer beam divider prevents further

This content downloaded from 128.141.48.10 on Mon, 5 Aug 2013 06:00:09 AM

All use subject to JSTOR Terms and Conditions
 waves
Free-space velocity of electromaggnetic 115
reflexion of this returned energy from the main beam divider. Thus, it is apparent
that when the c.p.i. pistons are so adjusted as to give the maximum transmitted
output, a rotation of the c.p.i. micrometer in either direction will cause a reduction
of output energy. When the movable carriage of the principal interferometer was
in the 'gauges inserted' position (i.e. set on the 971st minimum) and the ampli-
tude of the waves picked up by the receiving horns was balanced by a clockwise
rotation of the c.p.i., the c.p.i. micrometer was said to be at 'position A'. Anti-
clockwise rotation to produce amplitude balance led to 'position B'. If the c.p.i.
micrometer was moved from one position of maximum transmission to the next
the phase of its output was changed by 7T so that two new positions (A1 and B1)
of the micrometer were now possible, for equal output, when the device was used
for less than maximum transmission.

4. THE INHERENT ERRORS OF THIE MICROWAVE SYSTEM

All errors were of the same form, but greatly reduced in magnitude, as those
detailed in the description of the prototype interferometer. Apart from the effect
of diffraction, which reduces with increasing transmitting-receiving horn separation
and can be precisely eliminated (? 6), errors arise from: (1) multiple reflexion of
microwave energy between transmitting and receiving horns; (2) transmitted
energy scattered into the receiving horns from some fixed external surface such as
the floor; (3) errors inherent in the measurement of the refractive index of the
atmosphere, and thus in the reduction of wavelength measurements made in air to
the vacuum condition; (4) errors due to imperfections in the c.p.i.; (5) the errors,
already discussed, due to possible differences between the geometrical and micro-
wave axes of the four horns.
Error (1) depends only upon the separation between the transmitting and re-
ceiving horns and reverses for a 'A displacement of the receiving carriage. Thus,
this error was eliminated by making at least two wavelength measurements, the
carriage for the second being displaced by 1A from its initial setting.
The disturbance due to the second effect, always less than 1 part in 106, was
eliminated by making a number of wavelength measurements for slightly differing
transmitting-receiving horn separations, the transmitting horns being set in steps
a total of a few wavelengths either side of the mean position selected for a given
diffraction effect. Thus, the phase of the direct transmission to the receiving horns
was varied relative to scattered radiation without in any way upsetting the
evaluation of diffraction.
The estimated accuracy of the refractive index measurement was + 1.1 parts
in 101. Great care had to be exercised in order to ensure equilibrium between the
water vapour in the air sample passing through the refractometer and water
atoms occluded by the cavity walls. It was necessary to circulate the air for at least
2 h before making a refractive index measurement (i.e. before evacuating the
cavity resonator and noting the change of resonant frequency); thus only one
determination could be made at the middle of the period occupied by each wave-
length measurement. Actually, the interferometer could only be used on cloudy
8-2

This content downloaded from 128.141.48.10 on Mon, 5 Aug 2013 06:00:09 AM

All use subject to JSTOR Terms and Conditions
116 K. D. Froome
dry days, or days of constant drizzle, when the water vapour content of the air was
very stable.
When used in the manner described in the next section no error arising from
the c.p.i. could be found, even when deliberately maladjusted. Remembering the
success of the c.p.i. used at 24 Gc/s, this result is not surprising, for the precision
of manufacture of the 72 Gc/s c.p.i. was two orders of magnitude better. The
waveguide portions of the 72 Gc/s device were constructed from optically flat
stainless-steel pieces and it was estimated that the internal dimensions were uniform
to within + 1 juin. over the distance of the piston movements.

5. EXPERIMENTAL PROCEDURE

In order to make wavelength measurements free from all disturbing influences,

the following procedure was adopted.
(1) The observer, after having performed a preliminary run to adjust (by means
of the slip gauges) the 2 m combination end-standard to be closely equal to 970 half-
waves, set the movable receiving carriage in contact with the sensing anvil of the
Electrolimit gauge mounted on the 'honeycomb' wall.
(2) The c.p.i. having been set at a position of maximum transmission (the mean
of the A and B positions), the variable attenuator and phase-shifter (see figure 1)
were adjusted to give zero-detector current at the position of the first minimum.
(3) The temperature of the end-bar assembly was taken.
(4) The observer made four settings on the first minimum, reading the Electro-
limit gauge each time.
(5) The receiving carriage was run to the limit of its movement and the end-
bar assembly lowered into position between the Electrolimit head and the fiduciary
ball on the carriage.
(6) The c.p.i. was turned to position A and the observer noted the Electro-
limit gauge reading for four settings on the 971st minimum.
(7) Operation (6) was repeated for the position B of the c.p.i.; the mean of the
Electrolimit readings for the A and B positions being taken as the true position of
the 971st minimum.
(8) The length-standard was removed and operation (4) repeated to correct for
drift, after which the temperature of the length-bar was again taken.
(9) The c.p.i. micrometer was moved to the next position of full transmission
(defined as the mean of the Al and Bl positions) and the Electrolimit head moved in
its clamp by IA so that the position of the 1st minimum was again within the range
of the Electrolimit gauge.
(10) Operations (3) to (8) were repeated using the c.p.i. in the A' and B' posi-
tions instead of the A and B positions.
By taking the mean of the two wavelength measurements so far made, the effects
of multiple reflexions between transmitting and receiving horns were eliminated,
together with all the c.p.i. errors except one. The operations were continued:
(11) The refractive index of the air was determined by shutting off the slow air
circulation through the refractometer cavity and measuring the change of fre-
quency upon evacuation.

This content downloaded from 128.141.48.10 on Mon, 5 Aug 2013 06:00:09 AM

All use subject to JSTOR Terms and Conditions
 Free-space velocity of electromagnetic waves 117
(12) A quarter-wave spacer was inserted in the waveguide of the output arm
of the c.p.i. The two receiving horns were rotated through 180 about their (hori-
zontal) geometrical axes, thereby reversing the effect of any difference between
microwave and geometrical axis.
(13) Operations (1) to (10) were repeated. The mean of all the four wavelength
measurements thus made was then free from all errors arising from the microwave
system, with the exception of that due to diffraction and scatter from fixed
objects.
(14) The effect of scatter was removed by the method described in the previous
section.
(15) To determine and eliminate the effect of diffraction, fully corrected (opera-
tions (1) to (14)) wavelength measurements were made for seven different trans-
mitting-receiving horn separations. The actual experiments took place between
January and July 1956.

6. THE DIFFRACTION CORRECTION

Since the full analysis of the diffraction correction has already been described
(Froome 1954), only the results of it will be given here. As before, experiments
were so planned that the transmitted wavefronts, by the time they reached the
receiving horns, were nearly true spheres. If z is the distance separating one trans-
mitting horn aperture from its corresponding receiver the effect of diffraction is to
cause an addition retardation of the received wave (over and above z) by D, where

z Al + A2
2
A A4
Z3 + 4+ (1)

The constants A1, A2, etc., are a function of horn-aperture field distributions
and are independent of z. For plane wavefronts at the mouth of the tranismitting
horns and plane receiving surfaces the even-order terms A2,A4, etc., vanish and
the series converges rapidly. When phase errors are present at the horn apertures
the convergence is somewhat slower and the experiments must be planned so that
orders higher than 1/z3 are negligible. The first term of the series corresponds to
the truly spherical received wavefront condition and dominated all the other
terms for the experiments with the four-horn interferometer. The method of
evaluating the diffraction correction is such that the greater the ratio of the
A1/z term to all the other terms in the series, the less need be known about the
actual aperture field distributions.
The interference equation may be written
-INA=A-(DZ1-DZ2), (2)
where N is the number of half-waves (970) in the carriage displacement, Az.
(AZ = Z2- Z1) z1 and Z2 are the transmitter-receiver separations for each end of
the carriage displacement. (The carriage was disposed symmetrically between
the transmitting horns.) A is the true free-space wavelength in air.
In order to minimize the effect of uncertainty in the knowledge of horn aperture
distributions, (2) is best written

This content downloaded from 128.141.48.10 on Mon, 5 Aug 2013 06:00:09 AM

All use subject to JSTOR Terms and Conditions
118 K. D. Froome
where K is a constant (ideallyunity) determinedby a least squares solution of the
experimental observations. In terms of phase velocity in vacuo, equation (3)
becomes
co= c-2Knf(Dz, -DZ2)/N; (4)
co is the true free-space phase velocity and c the apparent (measured) velocity
(2nfAz/N). n is the measured refractive index of the air and f is the measured
microwavefrequency.
The results were evaluated from a least squares solution of equation (4) giving
both c0 and K. If the assumptions concerning horn aperture distributions are
roughly correct, then K comes out nearly equal to unity and the 'best' value of
co is obtained. For the ideal 'spherical wave' case where only the first term in
equation (1) is significant, equation (4) gives the same result for co regardless
of the nature of the assumed aperture distributions and regardless of the value
of K. The experimentswith the 72 Gc/s interferometerwere planned so as to make
the derived value of co very insensitive to the exact nature of the horn aperture
distributions (i.e. very insensitive to K).
For ideal rectangular-section horns propagating the Ho, mode and free from
aperturephase errorsthe theoretical form of Dz is
+
Dz= [O 18943a2 + lb20?0253A2]/z
10[4a
[0- +0289b4 +?0063a2b2 + (0 1795a6+?06127b6)/A2]/z3
+ O( l/Z5) + . . ., (5)

where the horn aperture dimensions are 2a in the H-plane and 2b in the E-plane.
If smoothly varying aperture phase errors (zero on horn axis) of magnitude
Pa2, Qb2(cm) at the horniedges are introduced, D, becomes
DZ= (l/k) [Afl/z - A2f82/z2-A3183/z3
+ O(L/z4) +***] (6)
where kl =2/A,
Ih=4(XI + U1),
(2/if)(Y1+V1) + 4(Y2+ V2+ 2XYl + 2U1V1)
= (2/7T)[(3X2+ 5X1-5 L)+ (3U2+ 5U21 512) +4(X1U1-Y1Vi)]
+ 8[1X3 + 3X1X - 3Y1Y2 + IOX,Y,2- OX3/3
+K3U3+ 3U1U2-3Vy2+10U1712--IU13U3],
with X1 = a2l(02976-0 0227p2-00004p4)
X =aa4(O 1942 - 0308p2 - 0.0002p4),
= 1662- 0 -380p2-O.OOlp4),
a6(0.

Y3 a2(O-1057p- 00020p3-0O0002p5),
Y2= a4(0 1084p-OsOO43p3-00002p5),
U1= b(O5236- 0*0328q2- 0-0033q-O0003q6),
U2 - - 0061 8q2- 00052q4- 0 0003q6),
b4(0O4935
U3 = b6(05537- 0 0957q2- 00073q4 - 00004q6),
V1= b2(02193q + 0O0069q3),
V2-b4(0O2953q+ 0-0059q3- 00003q5- -0001lq7),

This content downloaded from 128.141.48.10 on Mon, 5 Aug 2013 06:00:09 AM

All use subject to JSTOR Terms and Conditions
 Free-space velocity of electromagneticwaves 119
and p = 2kPa2/ff, q = 2/cQb2/
1,
a-a/A, b =bA.
(The series converge for p, q, : 1, i.e. for phase errors of not more than 2r at
the horn aperture edges. A spherical form of phase error is given by P = Q.)

7. RESULTS
Table 1 gives the measured phase velocities for the seven different positions of
the transmitting horns.
Table 2 shows the corresponding values of co after removal of the diffraction
effect.

TABLE 1. MEASURED VELOCITIES (REDUCED TO VACUUM VALUES)

separation of
transmitting and
receiving horns with
movable carriage in measured
centre of displacement phase velocity
(cm) (km/s)
629-5 299796 020
751-5 4*981
875-0 4-283
999.0 3 802
1120 5 3*583
1247 5 3-482
1367 5 3 259

TABLE 2. FINAL VALUES OF CO

horn separation (cm) co (km/s)
629-5 299 792*513
751-5 2 529
875-0 2*476
999.0 2-414
1120-5 2-478
1247-5 2-588
1367-5 2-512
mean co 299 792-501 + 0 059 [s.d.]

It is seen that the standard deviation of a single determination of velocity at any

particular value of the horn separation is just under 2 parts in 107. It should be
noted that the variation of the co results in table 2 includes, in addition to the un-
certainty of setting on a minimum, the random uncertainties of the air refracto-
meter and the temperatures of the length-bar and the air.
As with the other microwave interferometers (Froome I952, I954) it was found
that the actual diffraction correction was less than the purely theoretical value. That
is, insertion of the geometrical horn apertures (a -4 cm, b = 3 cm) in equation (5)
gave a value of K less than unity when the results were evaluated using equa-
tion (4), although the, value of co obtained was the same as that finally accepted.
The values of co in table 2 have been derived by putting a = 36 cm, b = 26 cm

This content downloaded from 128.141.48.10 on Mon, 5 Aug 2013 06:00:09 AM

All use subject to JSTOR Terms and Conditions
120 K. D. Froome
in equation (5) thereby obtaining from equation (4) a value of K closely equal to
unity. A small allowance has also been made for the effect of possible aperture
phase errors. This was done by substituting in equation (4) D, values obtained
from equation (6) on the assumption of bad phase errors near the rim of the horn
apertures. On the basis of this investigation it was decided to reduce the c0 values
obtained for supposed perfect plane waveforms (i.e. equation (5)) by 0-02 km/s
with an uncertainty of + 0.01 km/s. Table 2 incorporates this additional correction.
Table 3 lists the estimates of the possible systematic errors present in the
determination.
TABLE 3. SYSTEMATIC ERRORS
magnitude
expressed in
source of error km/s
length measurement ( + 2 in 107) + 0 060
refractive index of air ( + 1 1 in 107) + 0 033
length-bar temperature (? 0 006 ?C) + 0-020
air temperature (? 0 03 00) + 0 010
c.p.i. residual + 0-015
diffraction residual + 0 010
frequency ( ? 1 in 108) + 0-003

The standard deviation of a single determination (table 2) is increased to

010 km/s when modified by statistical combination with the estimated systematic
errors (table 3), so the result is

co= 299 792*50 + 0 10 km/s.

8. DiscussIoN
When the same value is assigned to the part of the length-bar combination used
for the experiment described herein and with the earlier equipment, the value for
co obtained with the prototype becomes 299 792-75 ? 03 km/s. In view of this
agreement there can be no doubt that the methods used with the prototype for
eliminating the disturbing effects of diffraction, scatter, etc., must have been very
satisfactory-in particular the accuracy of assessment of diffraction must have
been excellent, for corrections of up to 100 km/s were applied at the smallest
transmitting-receiving horn separations. Since the 72 Gc/s interferometer has
been used in an identical manner, but with a figure of improvement for those
effects of more than one order of magnitude, the estimated accuracy of this inter-
ferometer should also be correct.
The estimated accuracy of the determination of the length-bar combination
refers to interferometric measurement in terms of the red line of the cadmium
spectrum as emitted by the standard Michelson lamp running at the specified
conditions. The most recent information placed before the Consultative Committee
for the Definition of the Metre, in September 1957, shows that this source is re-
producible to 1 part in 107, or slightly better, and this is considered to be the
accuracy of determination of the length of the 1 m X-seetion end-bar. The ad-
ditional 1 part in 107 given in table 3 is due to further uncertainties arising from

This content downloaded from 128.141.48.10 on Mon, 5 Aug 2013 06:00:09 AM

All use subject to JSTOR Terms and Conditions
 Free-space velocity of electromagnetic waves 121
the comparisonsof the two 1 m cylindrical section end-gauges against the X-bar,
and in the measurementof the slip gauges and the 6 mm steel ball which were also
a part of the final form of the length standard.
In 1960 it is hoped to replacethe International Prototype Metre by a light-wave
standard, the Consultative Committee having already agreed to recommend the
adoption of the orange-red line of krypton-86 (Ava,c- 0605 78021 ) emitted by
a specially cooled lamp. The actual value of this wavelength has been obtained
from the mean of a number of comparisons by various standards laboratories
against the red line of cadmium (Avac= 0 644 02491 ) as emitted by the same
form of lamp used for the length measurements connected with the new velocity
determination. In any case,whatever wavelength standard is finally adopted the
wavelengthof the cadmiumline will not be altered.
The new value is in good agreementwith other recent velocity measurementsof
high precision, the most recent and readily available discussion of these being
that of Mulligan& McDonald(I957) and Bergstrand(I957). By means of his
Geodimeter,Bergstrand has made the most precise determinations of the velocity
of light using geodetic bases of the order of 10 km in length. In this latest paper
he summarizes all the Geodimeter results and gives two values for co, namely,
299792*75+ 0*34km/s and 299792*85+ 0.16 km/s. In each case the value quoted
incorporatesthe weighted mean of results from a number of bases, the second set
being obtained with an improved form of apparatus. For this latter set he states
the estimated accuracyof a determinationfrom a single base-line (itself the weighted
mean of a number of velocity measurements)to be about ? 0 4 km/s of which half
lies in the uncertainty of length measurement. It is also in excellent agreement
with the other N.P.L. determinations using microwaves: Essen's (I950) cavity
resonator gave co= 299 792*5+ 1 km/s; Froome's microwave interferometersgave
299792*6+0*7 (I952) and 299792*75+0-3 (I954, corrected I958). The microwave
determinations of the velocity of electromagnetic waves up to 1956 have been
reviewed by Froome (1956). Of interest also is the recent (I957) decision of the
Xllth General Assembly of the International Scientific Radio Union (U.R.S.I.)
to recommend the adoption of co= 299 792*5 + 0-4 km/s. This value was also ac-
cepted by the International Union for Geodesy and Geophysics (I.U.G.G.).

The author wishes to acknowledge the very great assistance received from the
following sources: The Royal Radar Establishment for the Pound microwave
frequency stabilizers and many excellent component designs; Mr C. W. Clenshaw
of the Mathematics Division (N.P.L.) for evaluating diffraction integrals; Dr L.
Essen for the provision of the frequency standard and many helpful suggestions;
Mr D. C. Barnes for designing the receiving carriage;MrsP. Anderton who made
the actual wavelenth measurements and assisted in many other aspects of the
determination; the MetrologyDivision workshop for the willing co-operation and
great care in construction of all the apparatus, particularly the constant-phase
interferometerand other waveguide components.
The work describedforms part of the researchprogramof the Nationa:lPhysical
Laboratoryand is published by permissionof the Director of the Laboratory.

This content downloaded from 128.141.48.10 on Mon, 5 Aug 2013 06:00:09 AM

All use subject to JSTOR Terms and Conditions
122 K. D. Froome

REFERENCES
Bergstrand, E. 1957 Ann. fran9. Chronom. 11, 97.
Essen, L. 1950 Proc. Roy. Soc. A, 204, 260.
Essen, L. 1953 Proc. Phy8. Soc. B, 66, 189.
Essen, L. & Froome, K. D. I95I Proc. Phy8. Soc. B, 64, 862.
Froome, K. D. 1952 Proc. Roy. Soc. A, 213, 123.
Froome, K. D. 1954 Proc. Roy. Soc. A, 223, 195.
Froome, K. D. 1955 Proc. Phy8. Soc. B, 68, 883.
Froome, K. D. 1956 J. Brit. Instn. Radio Engr8, 16, 497.
Froome, K. D. 1958 Nature, Lond. 181, 258.
Mulligan, J. F. & McDonald, D. F. 1957 Amer. J. Phy8. 25, 180.
Sears, J. E. & Barrell, H. 1932 Phil. Tran8. A, 231, 75.

This content downloaded from 128.141.48.10 on Mon, 5 Aug 2013 06:00:09 AM

All use subject to JSTOR Terms and Conditions