44
251 September 6 1974
Swallow'· has reported the results of a series of current
measurements made during a period of 14 months in the
western North Atlantic. The observed structure of the
currents may be interpreted in terms of bottom trapped
topographic Rossby wave modes", with velocity decaying
on a vertical scale of several kilometres, from a near
bottom maximum of roughly 10 cm s-'. The presence of
significant topographic structuring on horizontal scales of
10 km and smaller" should introduce further variability in
the basic current field in the lowest kilometre. The inter-
action of this topographically modified quasisteady current
field with the smaller scale topographical features is
responsible for the generation of internal wave energy. As
the internal waves propagate away from the bottom, the
relative frequency, kU, decays with height. That has
serious implications as far as the internal wave field is
concerned. The ultimate catastrophe for any given wave
component occurs when it encounters a critical level"'"
at which the relative frequency is reduced to the local
inertial frequency, f. Nonlinear effects inevitably destroy
the internal wave component in the vicinity of such a
critical level, so that as a consequence of the general decay
of background current speed with distance above the
bottom, a substantial portion of the internal wave energy
should be dissipated in the lowest kilometre or so of the
ocean.
The rate of production of internal wave energy may be
estimated for conditions representative of the western
North Allantic. A typical deep current speed is, U - 10
cm s-'. A representative value for the Brunt-VaisaIa
frequency is" N - 7 X 10- 4 s-'. Analysis of a large number
of topographical profiles of abyssal hills in the eastern
North Pacific· suggests that a value of H2 - 103 m' is
representative, and consistent with the results of an analysis
of sonic sections" in the North Atlantic. At a latitude of
30· N (f - 7 X 10-' s-'), then, E - 10 erg cm- 2 s-'. If it
is assumed that 50% of that energy goes into mixing in the
lowest kilometre of the ocean, then there must be an
associated buoyancy flux, F, - 5 X 10-' cm 2 s-'. With a
Brunt-Vaisiila frequency, N, - 7 X 10-< s-', that implies a
vertical mixing coefficient (A = F / N') of - 100 cm 2 S-1.
That is precisely of the same order as the vertical mixing
coefficient based on the analysis of observations of the
western North Atlantic nepheloid layer2. Further support
of the proposed relationship between mixing induce'd by
internal waves and the nepheloid layer is provided by the
apparent correlation between bottom roughness and thick-
ness of the nepheloid layer'. The mixing induced by internal
waves discussed here must be balanced in the overall
thermohaline energy budget. Acting over 50% of the North
Atlantic, the vertical mixing estimate made here implies a
rate of energy conversion of the order of IOU erg s-'. In
this connection, Wright17 has investigated the rate at which
available potential energy is released by the sinking of cold
water masses in the northern North Atlantic. He found
that the two Norwegian Sea overflows, and winter con-
vection in the Labrador Sea, contributed - 0.85 X IOU
erg s-'. The North Atlantic seems to be somewhat unique
in the intensity of its deep currents. Near-bottom currents
of 3 or 4 cm s-' are probably more representative of the
world ocean., so that an estimate of E, - 1 erg cm- 2 s-"
with a corresponding mixing coefficient, A, of the order of
10 em' s-' may be considered as more representative of the
world ocean.
THOMAS H. BELL, JUN.
Physical Oceanography Branch,
Ocean Sciences Division,
Naval Research Laboratory,
Washington, DC, 20375
Received April 19; revised July 12, 1974.
© 1974 Nature Publishing Group
Nature Vol.
, Eittreim, S., Gordon, A. L., Ewing, M., Thorndike, E. M.,
and Burc~hausen, P., in Studies in Physical Oceanography,
vol. 2 (edIt. by Gordon, A. L.), 19 (Gordon and Breach, New
York, 1972).
2 Eittreim, S., and Ewing, M., ibid., 123.
3 Connary, S. D., and Ewin.g, M., ibid., 169.
• Wimbush, M., and Munk, W., in The Sea, 4, part 1 (edit.
, by Maxwell, A. E.), 731 (lnterscience, New York, 1970).
Sverdrup, H. U., Johnson, M. W., and Fleming, R. H., The
• Oceans, Table 65 (Prentice Hall, Englewood Cliffs, 1942).
Munk, W. H., Deep Sea Res., 13, 707 (1966).
7 Garrett, C., and Munk, W., Deep Sea Res., 19, 823 (1972).
; BeI.I, T. H., thesis, Johns Hopkins Univ., Baltimore (1973).
Rhmes, P., Boundary Layer Meteorol., 4, 345 (1973).
,. Swallow, J. C., Phil. Trans. R. Soc., A270, 451 (1971).
11 Rhines, P., ibid., 461.
12 Swallow, ibid., 452.
" Booker, J. R., and Bretherton, F. P., J. Fluid Mech. 27, 513
(1967). '
14 Jones, W. L., J. Fluid Mech., 30,439 (1967).
" Reid, J. L., and Lynn, R. J., Deep Sea Res., 18, 1063 (1971).
16 Cox, c., and Sandstrom, H., J. oceanogr. Soc. Japan 20, 499
(1962). '
17 Wright, C. R., Deep Sea Res., 19, 865 (1972).
Thermonuclear micro-explosions
with intense ion beams
HIGH density pellet compression may be important in laser
fusion ' . In this way the minimum energy input for break-
even can be substantially reduced to levels within reach of
expected future laser technology. Further 2 when this concept
is applied to a fissionable pellet surrounded by a small neutron
reflector, preferably TD, which is compressed together with
the reflector, to high densities, very small critical masses
become possible; in addition the fission process could ignite
a thermonuclear reaction in the TD reflector. In these proposals
very large laser energies of ~ 1 MJ, which may be difficult to
reach, are required. As an alternative to laser fusion I have
proposed the use of intense relativistic electron beams 3 , but
here the beam pulse length is longer than for laser beams,
and the electron stopping range may be too long.
I present here an alternative idea based on the possibility
of generating intense ion beams in a magnetically insulated
diode4 • 5 • In such a diode a pulsed laser beam is directed on to
the anode surface as a high voltage is applied to the diode,
and a magnetic field directed parallel to the cathode surface
prevents the flow of electrons from the cathode to the anode,
but not the flow of ions from the anode to the cathode, pro-
vided that the magnetic field is strong enough to make the
electron Larmor radius, but not the ion Larmor radius, small
compared with the anode-cathode gap. The laser beam hitting
the anode surface will produce a surface plasma from which
the ions are drawn and accelerated towards the cathode. After
their passage through the cathode, which has the form of a
grid to minimise the energy losses by collision, the ions are
projected into a drift tube. This drift tube is filled with a
tenuous plasma of high electrical conductivity. Because of their
large inertia the ions draw into their flow electrons which
pass through the highly conducting plasma from the cathode
surface or the inner surface of the drift tube. They thus com-
pensate for an electric space charge field which would other-
wise build up. Furthermore, because of their small mass and
high mobility, the electrons will follow the ion flow with the
same velocity, thus also cancelling the self-magnetic field of
the ion beam entering the drift tube. The ion beam therefore
drifts freely as an electrically and magnetically neutral beam.
Calculations for magnetically insulated diodes pulsed with
several megavolt predict current densities of the order j - 3
kA cm- 2 such that for an area of - 30 cm 2 a total current of
I ~ 105 A could be produced 5 • To deliver a total energy of
~ 1 MJ with a voltage of V-I 07 V and current of I - 105 A
the total discharge time would be t - 10-6 s. In order to take
full advantage of the low ion velocities we may choose singly
Nature V ol. 251 September 6 1974
R~====~~====~~------~
_ _-+---2
r.I--------------4-~
Fig. 1 Time focusing of the beam.
ionised heavy ions, such as ions of lead or uranium, with
which the anode surface would have to be made or coated.
For a voltage Vand atomic weight A, the ion velocity is given
by v ~ 1.38xI06 Y (V/A) cm S-I. Putting A=200givesv=
105 y Vcm S - I. For V = 106 Y, V ~ 10 8 cm S - 1 and for V =
107 Y, V ~ 3 X IO H cm s -1. J propose that this variation of
the ion velocity with the voltage may be used to focus the
beam by increasing the voltage at the diode during the pulse.
Assume for example, that the pulse lasts ~ 0.7 x JO-6 sand
that an ion at the beginning of the pulse has a velocity of
:::= 10 6 cm s - I, as compared with another ion at the end of the
pulse with a velocity of ~ 3 x 10 8 cm s - 1. One can then easily
calculate that after the time :::= 10-6 s the ion emitted at the
end of the pulse would have caught up with the ion emitted at
the beginning of the pulse, at a distance of ~ 100 cm and
time ~ 3 x 10-7 s after the end of the pulse. So by shaping
the time behaviour of the voltage pulse properly at the diode
one can make ·the ions converge at the same time and place,
shortening the beam pulse and raising the beam power by
perhaps many orders of magnitude.
By making the ion source spherically symmetric and by
projecting a beam of spherical symmetry towards a common
centre of convergence, the total beam power density can be
further increased. This could be approximated by a large
number of diodes in a spherically symmetric arrangement,
with the diodes having a concave shape so that each beam
pulse would have the form of a spherical segment.
Figure 1 shows how the ion velocity would have to be
changed in time during the pulse to focus the beam at time
t = t o and at the common centre of convergence r = O. The
ions are emitted at the position r = R. The lines 1, 2 and 3
represent the r - t diagram for the initial and intermediate '
and the final ion of the pulse. Time focusing of the beam then
requires for the ion velocity to follow
4S
pulsed laser beam, two types of velocity spreads will occur:
first, by a differing degree of ionisation; second , by the initial
thermal motion. It is conceivable that by a properly chosen
laser frequency a high degree of uniform ionisation can be
achieved. The most important velocity spread therefore results
from the initial thermal ion energy at the anode surface. This
energy is of the order I eY. In the anode--<:athode gap the
ions are then accelerated by a voltage of - J06_10 7 Y, which
for singly ionised ions implies an energy of - J06_10 7 eY.
Therefore, the beam will be monoenergetic to the order - 10-6 to
10 - 7 corresponding to a relative velocity spread of - JO - 3 to
10 -3 ". Thus, a time focusing of the beam by the same factor
will be possible. This would imply, for example, that in the
case of the 100 MJ beam the energy could be delivered in a
time of about - 10-7 to 3 x 10 - 6 s, thus raising the beam power
to 3 x J015 W.
Because of the pulse shortening by time focusing which is
possible because of the low ion velocities, energy can be pumped
into the beam slowly by inductive energy storage devices
pulsing the diode. Since the peak voltage is several million
volts, this might be done with a cascade of pulse transformers.
Toe range of heavy ions in dense matter is rather small
(typically - 10-4 cm), so the beam can be easily stopped if
projected on to a solid piece of fusionable (or fission-fusion
hybrid) material. Furthermore, by properly shaping the time
behaviour of the laser light impinging on the anode, the ion
current and thus the beam power can be changed with time.
This is of importance if one wants to shape the beam pulse as
has been proposed by Nuckolls et aU for adiabatic high
density pellet compression with a power profile approximately
given by
E = const. (I - 1/1 0 ) - 15/6 (2)
It follows that the same power profile can be obtained with

v = R/Ct o - 1) (1)

If a target to be bombarded by the beam is placed at r = r 0,

the pulse length at the target " (at which r r 0) is then
shortened in comparison to the pulse length at the source
(at which r ~. R) by the factor, '/ , -- R/r o •
It follows that for an ion velocity of v and a pulse length ,
a linear beam can be focused in time after a distance x - V"t,
with a beam energy E ~ IVx/ v. For 1- 105 A, V - 10 7 Yand
v - ID· cm s -1 it follows that E - 10' x J. Therefore, for
x = 10 2 cm , with, - 10 - 6 S one would have E - 1 MJ and
for x = 104 cm with, - 10 - 4 s one would have E - 100 MJ.
A drift tube 100 m long is still well within the technical possi-
bilities and the energy of 100 MJ to be pumped into the beam
could probably be supplied by an inductive energy storage
system. But the energy stored in the drift tube can be delivered
in a much shorter time depending on the perfection of the time Fig. 2 Coalescence and projection of seve~al ion beams on to
focusing, which depends on the velocity spread of the ions. a thermonuclear pellet. I, Ion beam; D, drIft tube; P, thermo-
Since the ions are produced from the anode material by a nuclear pellet.

© 1974 Nature Publishing Group

46
the proposed ion beam technique. If the accelerating voltage
is V and the ion current I then
v = const. vi V (3)
E = IV (4)
From equations (1-4)
v = const. (1 - t/to)-I (5)
V = const. (1 - t/t o)-2 (6)
I = E/V = const. (1 - t/t o)1/8 (7)
Since the ion current is approximately proportional to the
laser light intensity, a simultaneous change in both the accele-
rating voltage and laser light intensity can produce the power
profile equation (2).
The proposed technique offers two possibilities for nuclear
micro-explosions:
First, the ignition of rather large micro-explosions with two
beams of ~ 100 MJ produced in two ~ 100 m drift tubes clash-
ing head on, compressing and heating between them a sand-
wich of dense thermonuclear or fission-fusion hybrid material.
For fusion pellets a minimum energy of ~ 10 2_10 3 MJ and
for hybrid fission-fusion pellets somewhat less energy would
be here required. It is therefore conceivable that two beams
each having an energy of ~ 100 MJ would suffice.
Second, the use of many beams approximating one beam of
spherical symmetry and which are projected simultaneously
into a spherical combustion chamber as shown in Fig. 2.
There, after having entered the chamber, the different beam
pulses would coalesce into one ion pulse having the shape of
a spherical shell, the thickness of which would decrease with
time as the pulse approaches the centre of convergence at
which the pellet to be ignited is placed. For reactor conditions
a total beam energy of ~ 1 MJ would be required, as in the
case of laser fusion, but since the focusing here may not be
as perfect as for optical radiation a larger input energy may
be needed; this does not, however, seem to present a problem
since the energy can be drawn from cheap inductive storage
devices. The pulse length required for solid pellet heating is
of the order ~ 10-9 s and less. If the pellet has the form of a
spherical shell the pulse can be somewhat longer. So if the
pulse is delivered from the diode in about ~ 10-6 s it has to
be compressed by a factor ~ 10 3 , shortening the pulse length
down to ~ 10-9 s. This implies a radial beam compression
from an initial beam width of ~ I m down to ~ I mm. It seems
that there is a fairly good chance that this can be done.
F. WINTERBERG
Desert Research Institute,
University of Nevada System,
Reno, Nevada 89507
Received April 22, revised June 20, 1974.
I Nuckolls, J., et al., Nature, 239,139 (1972).
2 Winterberg, F., Nature, 241, 449 (1973).
3 Winterberg, F., Phys. Rev., 174,212 (1968).
4 Winterberg, F., in Physics of High Energy Density, 370 (Academic
Press, New York, 1971).
, Sudan, R. N., and Lovelace, R. Y., Phys. Rev. Lett., 31, 1174
(1973).
Measurement of the speed of light
WE report here the completion of a determination of the
speed of light at the National Physical Laboratory'. The value
was obtained from the product of the measured frequency and
the wavelength, determined through up-conversion, of the
radiation from a CO 2 laser stabilised to the R(I2) transition of
CO 2 at 9.3 11m.
The frequency was determined relative to the caesium
standard, using successive stages of harmonic multiplication
and beat-frequency detection, with HCN and H 20 lasers used
as transfer oscillators. The value obtained 2 was v =
32,176,079,482 kHz, with a statistical uncertainty, CJ m , of
© 1974 Nature Publishing Group
Nature Vol. 251 September 6 1974
0.13 x 10 -9, and a systematic uncertainty of ± 0.7 x 10 -9.
(The systematic uncertainty is less than that published previously,
because only the reproducibility of the CO 2 stabilised frequency
is considered, rather than the uncertainty of obtaining the
unperturbed R(12) transition frequency.) The wavelength was
determined using interferometric measurements on up-
converted light at 0.68 11m and was related to the 66Kr primary
length standard, realised in accordance with the latest recom-
mendations of the Comite Consultatif pour la Definition du
Metre (CCDM). The value obtained 3 for the free space vacuum
wavelength is A. = 9,317,246.348 pm, with CJ m = 1.14 X 10-9
and a systematic uncertainty of ± 1.4 x 10 -9.
The product of these results gives the value for the speed of
light
c = 299,792,459.0 ± 0.8 m S-I
The uncertainty (corresponding to ± 2.7 x 10-9 ) is the arith-
metic sum of the standard error of the mean (1.1 x 10 -9) and
the total systematic uncertainty (± 1.6 x 10-9 ). Each of these
was derived by quadrature summation of the contributions
from the frequency and wavelength measurements.
Our measurement may be compared with values derived
from a frequency measurement, by Evenson et aU, of the
methane stabilised He-Ne laser at 3.39 11m, and the wave-
length measurements made in that and other laboratories. The
available data was reviewed by the CCDM in June 1973, when
the method of realising the metre through the 86Kr source was
reconsidered. It was decided that where asymmetry of the
spectral profile is observed, the defined wavelength applies to a
point midway between the peak and centre of gravity. Thus, the
resultofEvensonetal. becomes:c = 299,792,457.4 ± 1.1 ms- I •
On consideration of wavelength measurements in four labora-
tories, however, the CCDM recommended" for general use,
the value c = 299,792,458 m s-" with an uncertainty of
± 4 x 10-9 resulting from the uncertainties of the wavelength
measurements.
The satisfactory agreement of our result with those based
upon the 3.39 11m laser confirms both the recommended value
for c and the reliability of frequency and wavelength measure-
ment techniques involving infrared laser radiations.
T. G. BLANEY
C. C. BRADLEY
G. J. EDWARDS
B. W. JOLLIFFE
D. J. E. KNIGHT
W. R. C. ROWLEY
K. C. SHOTTON
P. T. WOODS
National Physical Laboratory,
Teddington, Middlesex, UK
Received May 17; revised July 2, 1974.
I Bradley, C. c., Edwards, G. J., Knight, D. J. E., Rowley, W. R. c.,
and Woods, P. T., Phys. Bull., 23,15 (1972).
2 Blaney, T. G., et al. Nature, 244, 504 (1973).
3 Jolliffe, B. W., Rowley, W. R. c., Shotton, K. c., Wallard, A. J.,
and Woods, P. T., Nature, 251, 46 (1974).
4 Evenson, K. M., et al., Phys. Rev. Lett., 29, 1346 (1972).
5 Terrien, J., Nouv. Revue Opt. 4, 215 (1973).
Accurate wavelength measurement
on up-converted CO2 laser radiation
W AYELENGTH measurements, which are part of an accurate
determination of the speed of Iight', have been made on the
radiation from a carbon dioxide laser, stabilised to the R(12)
transition at 9.3 11m by saturated fluorescence in an external
CO 2 cell, which has a measured frequency2 reproducible to
better than one part in 10". The problems of a direct inter-
comparison of infrared and visible wavelengths were avoided