The concentration dependence of the chemical shifts of the protons H-2, H-8 and H-I' of A T P and of Mg(ATP)'-, of all non-labile protons of adenosine, of H-5, H-6 and H-I' of UTP, and of H-5, H-6, H-l', and H-2' of uridine have been measured. The results for the purine derivatives are consistent with the isodesmic model of indefinite non-cooperative stacking; for adenosine K = 15For the pyrimidines, uridine and UTP, stacking is much weaker and the stability constant could only be estimated; for uridine K 2 0.5 M-l, and for UTP K z 0.3 M-'.Self-association of adenine [l] and purine [l], and of several purine nucleosides [2], including adenosine and inosine, has been observed by vapour pressure osmometry. It was concluded that indefinite noncooperative stacking occurred, and stability constants were calculated [2]; for adenosine K = 4.5 M-'. For pyrimidines, e.g. uridine, the interaction is very much weaker, and the calculated [l] stability constant is only 0.6 M-l. The NMR spectra of purine nucleosides are shifted in concentrated (0.1 M) solution, and this was attributed to the effect of the ring current in the stacked complexes ; however no significant shift was observed with the pyrimidine nucleosides uridine and thymidine [3] owing to the much smaller ring current in the pyrimidines [4]. In this early work external references were used; Bovey [5] repeated the experiment on purine using internal sodium trimethylsilylpropane sulphonate as reference, and concluded that, as no significant upfield shift of H-2 or H-6 is observed and as H-8 actually moves downfield, no stacking occurs. However sodium trimethylsilylpropane sulphonate is an unreliable reference in the presence of aromatic compounds [6,7], and thus, despite the use of an external reference, the earlier work is probably more reliable.An upfield shift due to stacking has also been observed [8-101 in NMR studies of adenosine 2: 3' and 5'monophosphates : the relaxation times of the protons H-2 and H-8, and the variation of the upfield chemical shift of the resonances H-1 ', H-2, and H-8 were said [9] to indicate 'head to tail' stacking, with the five-memAbbreviation. NMR, nuclear magnetic resonance. bered and six-membered rings alternating in the stack. These 'H NMR studies [8-101, a 'H NMR study [ll], ultracentrifuge studies 112,131 and a kinetic study [14] show that adenosine 5'-monophosphate [8 -121 and other adenine derivatives [13,14] do not merely dimerize, but form 'indefinite' stacks. Although difference ultraviolet spectroscopy [15] and difference circular dichroism [15,16] have been used to study the self-association of ATP, and the effect of pH, temperature, and of magnesium ions on this association, shifts in the 'H NMR spectrum of ATP, which would confirm that this association is due to stacking, have not been reported.In the course of a series of studies[l7-191 on the stacking in binary systems such as bipyridyl/ATP4-, phenanthroline/ATP4-, and bipyridyl/UTP4-, and in the ternary systems such as Zn2 /bipyridyl/ATP4-, Mg2 ' /phenanthrol...