802 Nature Vol.

299 28 October 1982

Received 15 June ; accepted I September 1982. 20. Kingsbury. R. & McKnight. S. L. Science 217.316- 324 (1982 ).
21. Larsen. A . & Weintraub. H . Cell 19,609-672 (1982).
22. Proudfoot. N. J. & Brownlee. G . G . Nature 163. 211-214 (1976).
1. Kagi. J. H. R. & Nordberg. M. (eds) Metallothiontin (Birkhauser. Basle. 1979). 23 .Calos. M. P. & Miller. J. H . Cell 20, 579-595 (1980).
2. Karin. M. & Herschman. H. R. Science 104. 176-177 (1979). 24. Hollis. F. G . et al. Nature 296, 321- 325 (1982 ).
3. Pulido. P .• Kalli. J. H. R. & Vallee. B. L. Biochemistry 5, 1768- 1777 (1966). 25 .Leuders. K .• Leder. A .• Leder. P. & Kufl. E . Naturt 195,426-428 (1982).
4. Rudd. C. J. & Herschmann. H. R. Tox. appl. Pharmac. 47, 273-278 (1979). 26. Van Arsdell. S. W. et al. Cell 16, 11-17 (1981 ).
5. Karin. M. & Henchman. H. R. Eur. J. Biocht rn . 107,395-401 (1980). 27. Jagadeeswaran. P .• Forget. B. G. & Weissman. S. M. Cell 16, 141-142 (1982).
6. Kissling. M. M. and Kagi. J. H. R. FEBS Lett. 81, 247-250 (1977). 28. Nishioka. Y .• Leder. A. & Leder. P. Proc. natn. A cad. Sci. U.S.A. 77, 2806-2809 (1980).
7. Karin. M. <I al. Nature 186, 295-297 (1980). 29. Wilde. C. D. et al. Nature 197,83- 84 (1982).
8. Karin. M .. Slater. E . P. & Herschman. H. R. J. all. Physiol. 106,63-74 (1981). 30. Shaul. Y .. Kaminichik. J. & Aviv. H. Eur. J. Biochem. 116, 461-466 (1981).
9. Durnam. D. M. & Palmiter. R. D. J. bioi. Chtm. 256, 5712-2716 (1981). 31. Perry. R. P. etal. Proc. natn. Acad. Sci. U.S.A. 77.1937- 1941 (1980).
10. Haaer. L. J. & Palmiter. R. D. Nature 191, 340-342 (1981). 32 .Hofer. E . & Darnel, J. E. Cell 13, 585-593 (1981).
11. Karin. M. & Richards. R. Nucleic Acids Res. 10,3165-3173 (1982). 33 .Bell. G .• Karam. J. H. & Rutter. W. J. Proc. natn. Acad. Sci. U.S.A. 78,5759-5763 (1981 ).
12. Lawn. R. M. "al. Cell IS, 1157-1174 (1978 ). 34. Rigby. P. W. J. et al. J. molte. Bioi. 113, 237-251 (1977).
13. Southern. E. M. J. molec. Bioi. 98,503-517 (1975 ). 35 .Wahl. G . M., Stern, M. & Stark. G . R. Proc. natn. Acad. Sci. U.S.A. 76, 3683-3687 (1979).
14. Benton. W. D . & Davis. R. W. Science 196, 180-182 (1977). 36. Maxam. A. & Gilbert. W. Meth. Enz ym. 65, 499- 559 (1980).
15. Glanville. N.• Durnam, D . M. & Palmiter. R. D . Nature 191,267-269 (l 98\). 37. Sanller. F .• Nicklen. S. & Coulson. A. R. Proc. natn. Acad. Sci. U.S.A. 74. 5463-5468
16. Breathnach. R. et al. Proc. natn. A cad. Sci. U.S.A . 75, 4853-4857 (1978 ). (1979).
17. Weaver. R. F. & Weissman. C. Nucleic A cids Res. S. 1175-1193 (1979). 38. Goodman, H. M. Meth. Enz ym. 65, 63-64 (1980).
18. Kayb. K. E .• Warren. R. & Palmiter. R. D. Ct1l19. 99-108 (1982). 39. Heidecker. G .• Messing. J. & GronenOOrn. B. Gene 10.69-73 (1980).
19. Brinster. R. L. et al. Nature 296.39-42 (1982). 40. O'Farrel. P. Focus 3.1-3 (l98\).

LETTERS TO NATURE
A single quantum cannot be cloned on an incoming photon with polarization state Is):
IAo)ls)-+ IA.)lss) (1)
W. K. Wootters·
Here lAo) is the 'ready' state of the apparatus, and lAs) is its
Center for Theoretical Physics. The University of Texas at Austin, final state, which mayor may not depend on the polarization
Austin, Texas 78712, USA of the original photon. The symbol Iss) refers to the state of
the radiation field in which there are two photons each having
W. H. Zurek the polarization Is). Let us suppose that such an amplification
can in fact be accomplished for the vertical polarization It)
Theoretical Astrophysics 130-33, California Institute of Technology, and for the horizontal polarization 1-). That is,
Pasadena, California 91125, USA
(2)
and
If a photon of definite polarization encounters an excited atom, (3)
there is typically some non vanishing probability that the atom
wiD emit a second photon by stimulated emission. Such a photon According to quantum mechanics this transformation should
is guaranteed to have the same polarization as the original be representable by a linear (in fact unitary) operator. It there-
photon. But is it possible by this or any other process to amplify fore follows that if the incoming photon has the polarization
a quantum state, that is, to produce several copies of a quantum given by the linear combination a It) + f31- )-for example, it
system (the polarized photon In the present case) each having could be linearly polarized in a direction 45 0 from the vertical,
the same state as the original? If it were, the amplifying process so that a = f3 = 2- t !2-the result of its interaction with the
could be used to ascertain the exact state of a quantum system: apparatus will be the superposition of equations (2) and (3):
in the case of a photon, one could determine Its polarization IAo)(al t )+f3I-» -+ aIAvert)ltt)+f3IAhor)I~) (4)
by first producing a beam of identically polarized copies and
then measuring the Stokes parameters·. We show here that the If the apparatus states IA v..t ) and IA hor) are not identical, then
linearity of quantum mechanics forbids such replication and the two photons emerging from the apparatus are in a mixed
that this conclusion holds for all quantum systems. state of polarization. If these apparatus states are identical,
Note that if photons could be cloned, a plausible argument then the two photons are in the pure state
could be made for the possibility of faster-than-light communi-
cation 2 • It is well known that for certain non-separably corre- a Itt) + f31~) (5)
lated Einstein-Podolsky-Rosen pairs of photons, once an In neither of these cases is the final state the same as the state
observer has made a polarization measurement (say, vertical with two photons both having the polarization al t )+f3I-).
versus horizontal) on one member of the pair, the other one, That state, the one which would be required if the apparatus
which may be far away, can be for all purposes of prediction were to be a perfect amplifier, can be written as
regarded as having the same polarization 3 • If this second photon
could be replicated and its precise polarization measured as
above, it would be possible to ascertain whether, for example,
the first photon had been subjected to a measurement of linear which is a pure state different from the one obtained above by
or circular polarization. In this way the first observer would be superposition [equation (5)].
able to transmit information faster than light by encoding his Thus no apparatus exists which will amplify an arbitrary
message into his choice of measurement. The actual impossibil- polarization. The above argument does not rule out the possibil-
ity of cloning photons, shown below, thus prohibits super- ity of a device which can amplify two special polarizations, such
luminal communication by this scheme. That such a scheme as vertical and horizontal. Indeed, any measuring device which
must fail for some reason despite the well-established existence distinguishes between these two polarizations, a Nicol prism
of long-range quantum correlations6 - 8 , is a general consequence for example, could be used to trigger such an amplification.
of quantum mechanics 9 . The same argument can be applied to any other kind of
A perfect amplifying device would have the following effect quantum system. As in the case of photons, linearity does not
forbid the amplification of any given state by a device designed
• Presen' address : Department of Physics and Astronomy. Williams College. Williamstown, especially for that state, but it does rule out the existence of a
Massachusetts 01267. USA. device capable of amplifying an arbitrary state.

© 1982 Nature Publishing Group

Nature Vol. 299 28 October 1982 803
Milonni (unpublished work) has shown that the process of quantum limits on the noise in amplifiers 10.11. Moreover, an
stimulated emission does not lead to quantum amplification, experiment devised to establish the extent to which polarization
because if there is stimulated emission there must also be-with of single photons can be replicated through the process of
equal probability in the case of one incoming photon-spon- stimulated emission is under way (A. Gozzini, personal com-
taneous emission, and the polarization of a spontaneously emit- munication; and see ref. 12). The quantum mechanical predic-
ted photon is entirely independent of the polarization of the tion is quite definite; for each perfect clone there is also one
original. randomly polarized, spontaneously emitted, photon.
It is conceivable that a more sophisticated amplifying We thank Alain Aspect, Carl Caves, Ron Dickman, Ted
apparatus could get around Milonni's argument. We have there- Jacobson, Peter Milonni, Marian Scully, Pierre Meystre, Don
fore presented the above simple argument, based on the Page and John Archibald Wheeler for enjoyable and stimulating
linearity of quantum mechanics, to show that no apparatus, discussions.
however complicated, can amplify an arbitrary polarization. This work was supported in part by the NSF (PHY 78-26592
We stress that the question of replicating individual photons and AST 79-22012-A1). W.H.Z. acknowledges a Richard
is of practical interest. It is obviously closely related to the Chace Tolman Fellowship.

Received 11 August; accepted 7 September 1982. 7. Fry. E. S. & Thompson. R. C. Phys. Rev. Lett. 37, 465-468 (1976).
1. Born. M. & Wolf, E. PrinCiples of Optics 4th edn (Pergamon, New York, 1970). 8. Aspect. A.. Grangier. P. & Roger. G. Phys. Rev. Lett. 47,460-463 (1981).
2. Herbert. N. Found. Phys. (in the press). 9. Bussey. P. J. Phys. Lett. 9OA, 9-12 (1982).
3. Einstein, A., Podolsky, B. & Rosen. N. Phys. Rev. 47, 777-780 (1935). 10. Haus. H. A. & Mullen, J. A. Phys. Rev. 128,2407-2410 (1962).
4. Bohm. D. Quantum Theory, 611-623 (Prentice· Hall, Englewood Cliffs. 1951). 11. Caves, C. M. Phys. Rev. D15, (in the press).
5. Kocher, C. A. & Commins. E. D. Phys. Rev. Lett. 18, 575-578 (1967). 12. Gozzini. A. Proc. Symp. on Wave-Particl€ Dualism (eds Diner, S., Fargue, D., Lochak.
6. Freedman. S. J. & Clauser, J. R. Phys. Rev. Lett. 28,938-941 (1972). G. & Selleri, F) (Reidel. Dordrecht, in the press).

The Crab Nebula's progenitor lower limit of the large helium-to-hydrogen ratio means that
at least half of the ejected material must have come from the
helium layer.
Ken'ichi Nomoto·, Warren M. Sparkst, Arnett (ref. 7 and refs therein) systematically evolved helium
Robert A. Fesen:j:, Theodore R. Gull:j:, S. Miyaji:j: cores of various masses (Ma) into late stages of evolution. He 1
& D. Sugimoto· compared Davidson's8 derived abundances of the Crab nebula
with calculated abundances from the Ma = 4.0 M 0 model, which
* Department of Earth Science and Astronomy, University of was his lowest-massed, highly evolved helium core (correspond-
Tokyo, College of General Education, 3-8-1 Komaba, Meguro, ing to approximately a 15 M _ star). Combining all the material
Tokyo 153, Japan above the helium-burning shell (his case B) with enough inter-
t Group X-5, Mail Stop F669, Los Alamos National Laboratory,
Los Alamos, New Mexico 87545, USA
stellar material to obtain X Hel X H = 8, he found good agreement
t Laboratory for Astronomy and Solar Physics, Goddard Space with XNIXHe and XOIXHe of Davidson's8 'model 1'. However,
Flight Center, Greenbelt, Maryland 20771, USA the calculated value of XCIXHe was too large by a factor of 3~.
At that time, the Crab's carbon abundance had not been directly
measured and Arnett suggested several possibilities: the
The study of supernovae is hampered by an insufficient knowl- inferred carbon abundance was too low, the carbon was hidden
edge of the initial stellar mass for individual supernova. Because in the filaments, or a lower-mass helium core, -3 M 0, was
of large uncertainties in estimating both the total mass of a more appropriate.
remnant (including the pulsar or black hole) and any mass loss Using recent UV observations with the International Ultra-
during the pre-supernova stages, the main sequence mass of violet Explorer, Davidson et ai. 5 have established that the
the progenitor cannot be accurately determined from observa- carbon abundance is nearly solar. They also showed that the
tions alone. To calculate an initial mass, one must rely on a hydrogen and helium seemed to be fairly well mixed and, as
combination of both theory and observation. Limits on the carbon is convectively mixed in the helium layer, this would
progenitor's mass range can be estimated by the presence of a argue against carbon being hidden in the filaments. However,
compact remnant and comparison of the observed nebular IR observations by Dennefeld and Andrillat 9 showed that the
chemical abundances with detailed evolutionary calculations·. strength of [C I] A 9,850 relative to [S III] A 9,069 varied with
The Crab Nebula is an excellent choice for investigation because position in the Crab. The strongest [C I] line would indicate a
it contains a unique combination of characteristics: a central rather large carbon abundance if the ionizing flux is constant.
neutron star (pulsar) and a bright, well studied nebula having Whether the IR observations indicate variation in the carbon
little or no swept-up interstellar material. In fact, several abundance, variation in the ionizing flux, or high densities in
studies·..... have suggested an initial mass of -10 M 0 for the neutral cores is not known. For the remainder of this report
Crab progenitor. Recently, Davidson et al.S, quoting two of us we will assume the carbon abundance as determined by David-
(K.N. and W.M.S.), state that the Crab's progenitor had a mass son et ai. s,
slightly larger than 8 M 0. Here we present in detail the reason- The existence of a pulsar in the Crab indicates that the
ing behind this statement and suggest the explosion mechanism. progenitor's mass was larger than the upper mass limit (8 ±
Briefly, the Crab consists of a pulsar (assumed here to have 1 M 0)10 for degenerate carbon ignition. Degenerate carbon
a mass of "'" 1.4 M 0) and a nebula mass of 1.2-3.0 M 0 (refs ignition results in carbon deflagration l l which completely dis-
5,6) which has a helium overabundance of 1.6<XHe /XH <8 rupts the star, leaving no compact remnant. Lower-mass stars
(where X is an element's mass fraction). The oxygen abundance that lose enough mass to avoid degenerate carbon burning
(Xo ) is -0.003 (refs 5,6), which is less than the solar value of eventually become white dwarfs. Stars massive enough (;,.8 M 0)
0.007, while the oxygen-to-hydrogen ratio is approximately to burn carbon non-degenerately will eventually undergo a core
solar. The carbon-to-oxygen ratio is 0.4 <XciXo < 1.1 (ref. 5). collapse initiated either by electron capture 12 onto Mg, Ne and
Nitrogen may be slightly overabundant, while neon, sulphur o or by burn-out of all the available fueI 13 •14 • When the collaps-
and iron abundances are uncertain but are probably not greatly ing core reaches neutron-star densities, stability is regained.
over- or underabundant. Because the Crab Nebula is helium- Although detailed calculations of the collapse remain inconclus-
rich but not oxygen-rich, the hydrogen-rich (solar abundances) ive, it is generally felt that the core will overshoot its equilibrium
envelope and the helium layer of the progenitor star were position and then rebound, initiating a shock wave 4 • This shock
ejected but the oxygen-rich layer below the helium layer was wave ejects the outer material but not the core, reSUlting in
not. The lower layers must have formed the neutron star. The both a supernova nebula and a pulsar 1s • In more massive stars

© 1982 Nature Publishing Group