Ion probe data are documented for zircons from tuffs within the early Llandovery, the mid-Caradoc and the Ludlow. Pb-206/U-238 ages for tuff magmatism have been interpreted using mixture-modelling to distinguish inheritance and Pb loss. Comparisons with the reference zircon SL13 have been improved through a direct determination of the component of secondary ion discrimination caused by changes in target potential. Interpretation of the SHRIMP data for the Birkhill ash (Scotland, Llandovery) is ambiguous. The more conservative possibility is that most zircons are 439 Ma, in close agreement with the previous isotope dilution ages for the same zircon concentrate. The other is that the 439 Ma group should be split into an inherited population at similar to 447 Ma, with a minority at similar to 434 Ma that corresponds with the ash volcanism. Although imprecise, the latter is detectably younger than the multi-grain MSID age, which itself might be a composite of the same two ages. Most zircon analyses from the mid Caradoc Pont-y-ceunant Ash, Wales, fit an age-group at 452.5 Ma, similar to its published Pb-206/U-238 age by MSID, with a definite older age group at similar to 476 Ma but none showing Pb loss. By contrast, those from the Millbrig bentonite (Virginia) of the same age mainly fall in a well-defined post-eruption age group at 435 Ma, while the remainder give 456 Ma. Most zircon analyses from the Kinnekulle bentonite, Sweden, fall into an apparent 464 Ma group which exceeds other estimates far the mid-Caradoc magmatism. It is interpreted to be a composite age, caused by an inability to resolve it into a younger magmatic and older inherited group owing to the larger analytical errors of the Kinnekulle data. The best SHRIMP estimate for the mid-Caradoc volcanism is 452.6+/-1.7 Ma found by combining the ages for the three volcanic units. During unmixing of the combined ages, the Kinnekulle ages are redistributed and the 464 Ma 'group' vanishes. Precambrian grains are present in all the above volcanics. The original and new zircon analyses from the Laidlaw Volcanics (Canberra, Australia) of Ludlow age, are dominated by three groups of inherited zircons at similar to 436 Ma, similar to 450 Ma and similar to 476 Ma, which makes it unfavourable for time-scale definition using zircons. The youngest zircon age group is 417.5 Ma (similar to 30%), but this is not associated with overgrowths on older grains or with wholly younger grains. Instead, it is composed of sporadic low ages within older grains suggestive of Ph loss rather than magmatic zircon growth. Nevertheless, the age for volcanism is 420.7+/-1.1 Ma based on published Rb-Sr and K-Ar dating, so that the youngest zircon group does appear to be associated with volcanism. One zircon U-Pb age for the Frasnian by MSID is much older than a precise age by other decay schemes, and another for the Lochkovian is significantly older than a recent SHRIMP age for the same Stage. By small changes in the common Pb composition, both MSID ages can be changed from single volcanic ages affected by Pb loss to an inherited and younger volcanic age, which removes the conflict with the other determinations. A zircon-based geological time-scale is constructed from the Ordovician to the Carboniferous using the time-points presented and discussed in Parts I and II of this paper.
