© Nature Publishing Group 1974 Nature Vol. 250 August 30 1974 of Physical Review Letters (32, I 336; I 974), following their demonstration of the phenomenon in Nature Ph ysical Science (235, 63; 1972). In their experiment they studied the light absorbed rather than emitted by atoms. The intense light from a tune- able wavelength laser is shone on the atoms and is absorbed by a set of atoms whose component of velocity in the direction of the light enables the Doppler-shifted wavelength to be matched to their absorption. A weaker probe beam from the same laser, and therefore having the same wavelength, made to travel in the opposite dir- ection so it will be ignored by this set of atoms but absorbed by another set whose members have an equal and op- posite velocity component. There is a special case when the wavelength is tuned to be such that the atoms which absorb the light are moving at right angles to the light paths and hence have a very small Doppler shift. Both sets •then comprise the same atoms and these being partially saturated by the intense beam, absorb less of the probe beam. It is this decrease in absorption which is detected as a nar- row signal without any appreciable Doppler width as the laser wavelength is tuned through the spectral line. There is a double advantage in all this. First , the line-broadening Doppler effect, which previously blurred to- gether the closely spaced fine structure components of the spectral line, is eliminated. Second, as this technique entails absorption of nearly mono- chromatic laser light the interfero- meter used to measure the wavelength can have a very small spectral range and a high resolving power; previously all the fine structure components were excited simultaneously in a lamp and the spectral range of the interferometer had to be large enough to encompass them all, with a consequent loss of resolution. Hanscih et al. measured <the absolute wavelengths of some fine structure components in the Balmer-a lines of hydrogen and deuterium to determine a value for the Rydberg constant, which characterises the wavelength of the spectral lines, and a value for the isotope shift between hydrogen and deuterium , which essentiaHy measures the ratio of the mass of the electron to that of the proton. The accuracy achieved was an order of magnitude greater than that obtained by the emission teahnique . Physicists in the past have used new techniques to measure accurately the spectrum of the hydrogen atom and in each successive instance the results have not agreed with the current theoretical description. Each conse- quent improvement of the theory has revolutionised physics-witness Bohr's concept of stationary states which led to quantum theory . And again, cooling hydrogen lamp in liquefied gases na rrowed the spectral lines and revealed the fine structure which was ultimately explained by Dirac's relativistic formu- lation of the quantum theory . Here the requirement of negative energy states suggested the existence of the positron and brought about the science of elementary particle physics. Further minute discrepancies (Lamb shifts), confirmed by another new technique of radio-frequency spectroscopy, resulted in the quantum theory of radiation fields and particles, quantum electro- dynamics (see Series, The Spectrum of Atomic Hydrogen; OXlford University Press 1957). Now measurements of in- creasing refinement can be anticipated using this latest technique of saturated absorption, particularly as what I have described is only one of a number of allied methods in which the natural resonances of atoms which give rise to their spectral lines can be explored by tuneable wavelength lasers. (One of these methods which is especially pro- mising is two-photon absorption which has recently been demon strated experi- mentally, see, for example, Physics Today, 27, 17; 1974.) Modern physics owes much to the exploration of the hydrogen atom at successively higher levels of precision and there is no reason to suppose that this process cannot continue. Missing link in folding of trypsin inhibitor from Barry Robson A PARALLEL could be drawn between the the folding of a globular protein and biological evolution in that both represent a transition from a state of disorder to one of order and function. Moreover, both processes have their missing links without which it is impos- sible to · verify hypotheses concerning mechanism. The difficulty of finding the mising links in the foldh1g of a glob- ular protein is because the process is, for most proteins , approximately two state, involving un stable and short- lived intermediates which cannot normally be isolated. Three articles by Creighton (1. molec. Bioi., 87, 563, 579, 603-624; 1974) describe a procedure for trapping and characterising such inter- mediates. As an experimental system, Creighton chooses bovine pancreatic trypsin in- hibitor, a small protein of only 58 amino acid residues which inhibits the catalytic function of certain other proteins, including trypsin and chymo- trypsin. The native structure of the inhibitor as obtained from living tis- 707 sue has been well characteri sed by X-r ay crystallographic analysis (Huber et a/ ., Naturwissenschaften, 57, 389- 392; 1970). As in the case of other globular proteins, the native structure is a compact, relatively rigid confor- mation maintained by non-covalent interactions (that is by van der Waal's, electrostatic, hydrogen bond and hydro- phobic interactions), as well as by cov- alent disulphide bridges between cys residues. Trysin inhibitor has three such disulphide bridges. Since this compact, biologically active structure is to be the end point of th e folding process, Creighton first unfolds the inhibitor in a concentrated solution of guanidinium chloride, which breaks the non-covalent inter- actions , and dithiothreitol, which breaks the disulphide bridges. In such condi- tions it is known that typical protein molecules assume highly flexible, open conformations which are, strictly speaking, not single states at all but a whole collection of conformational states of roughly equivaleillt energies separated by low conformational energy barriers . The importance of starting with this collection of states is that all the information for directing the folding process towards the native conforma- tion must initially reside only in the sequence of amii1o acid residues char- acteristic of trypsin inhibitor and ulti- mately coded for in the chemical structure of the gene. It is therefore regrettable vhat the conformational dis- order of the inhibitor was not actually proven before refolding, but only assumed by analogy to the behaviour of ofher proteins in these conditions. Al- though Creighton's indirect evidence for initial conformational randomness does not , however, exclude the poss- ibility of a fairly high degree of residual conformational structure, nobody would deny that the absence of such structure is a reasonable bet which one hopes will be verified by future hydro- dynamic tests. Creighton initiates refolding of this assumed random structure by removing the guanidinium chloride by dilution and adding a disulphide reagent such as oxidised dithiothreitol. The refolding reaction is then quenched at different times by the addition of iodoacetate or iodoacetamide which stop the further making and breaking of disulphide bridges. The species so trapped at various stages of refolding are sub- sequently analysed by acrylamide gel electrophoresis. When any remammg unbridged residues of the trapped intermediates are carboxymethylated, the relative mobilities of the intermediates in the acrylamide gel reveal that some con- tain one and some two disulphide bridges. Consideration of the time course of the appearance and disap-