Geol. Mag. 144 (4), 2007, pp. 633–642. c 2007 Cambridge University Press 633 doi:10.1017/S0016756807003391 First published online 6 June 2007 Printed in the United Kingdom Origins and assessment of snowball Earth hypotheses W. BRIAN HARLAND† Department of Earth Sciences, University of Cambridge, Cambridge CB2 3EQ, UK Note from the Editors – Brian Harland was for many years an editor of this journal. He was also a seminal figure in the origins of the current ‘snowball Earth’ debate, having recognized in 1964 the significance of coupling emerging palaeomagnetic data on palaeolatitude with his interpretations of diamictites. Harland worked extensively in the Arctic and knew well many of the workers involved in the arguments surrounding the origin of diamictites. He thus had a unique perspective on the evidence and the disputes surrounding it. This was his last paper but he was not able to complete it before he died. However, with the help of Professor Ian Fairchild to whom we are indebted, the editors have lightly revised this work which is presented as the personal view of one of the key figures with a very broad stratigraphic appreciation of the problems of ‘snowball Earth’. Abstract – Records of Precambrian glaciation onwards from the late nineteenth century led to the concept of one or more major ice ages. This concept was becoming well advanced by the mid 1930s, particularly through the compilation of Kulling in 1934. Even so tillite stratigraphy shows that glaciation was exceptional rather than typical of Earth history. Some Proterozoic tillites, sandwiched between warm marine facies, indicate low, even equatorial palaeolatitudes as determined magnetically, and more recently led to ideas of a snow- and ice-covered ‘snowball Earth’. However, interbedded non-glacial facies as well as thick tillite successions requiring abundant snowfall both militate against the hypothesis of extreme prolonged freezing temperatures referred to here as an ‘iceball Earth’ in which all oceans and seas were sealed in continuous ice cover. On the other hand tropical environments were interrupted by glaciation several times in the Proterozoic, something that did not recur in the Phanerozoic. The term ‘snowball Earth’ is consistent with the established view of extremely widespread Proterozoic glaciation, but the ‘iceball Earth’ version of this is not compatible with the geological record. Keywords: snowball Earth, iceball Earth, glaciation, tillite. 1. Introduction were sealed by ice. The latter interpretation is here referred to as ‘iceball Earth’. However, a less extreme Thoroughly tested palaeomagnetic evidence has es- snowball hypothesis entailing low latitude (tropical) tablished that late Proterozoic glaciation extended to glaciations is now well established and may be typical equatorial palaeolatitudes (Harland & Herod, 1975; for Precambrian rather than Phanerozoic ice ages. In the Evans, 2000). This led to the hypothesis of a ‘snowball following text, the term tillite, implying a sedimentary Earth’ by Kirschvink (1992) who also inferred that rock with a recognizable glacial component, terrestrial the planetary cooling was stabilized by an ice-albedo or marine, is used here as in most of the literature feedback mechanism. Hoffman et al. (1998) developed quoted. this concept further to make a more all-embracing theory, but at the same time developed a more extreme interpretation whereby all ocean surfaces were sealed 2. Origins of the hypothesis of global glaciation by ice over a time period of 4 to 30 million years. Snow- 2.a. British Isles ball Earth has attracted wide attention as a stimulating popular science topic (Walker, 1999, 2003; Hoffman & The first Precambrian glacial evidence to be so Schrag, 2000; Fairchild, 2001) as did its precursors interpreted was at Port Askaig in Islay, Scotland, by (Harland & Rudwick, 1964; Hambrey, 1992). In recent James Thomson (1871). It is appropriate that a most literature there has been a lack of emphasis on the thorough description of any glacial sequence was of this historical development of ideas, although this has been and related tillites (Spencer, 1971). Archibald Geikie partially corrected by Hoffman & Schrag (2002). This (1880) considered a Torridonian diamictite as possibly brief paper develops the historical element further glacial. James Croll in 1864 (1875) had suggested and shows that early literature had already established a genetic relationship of such an ice age with the the essential nature of the glaciation, but also had advent of Cambrian life, an idea that has recurred precluded the interpretation that all ocean surfaces and been progressively developed over time (Hicks, 1876; Ramsey, 1880; Coleman, 1926; Mawson, 1949; † deceased 1 November 2003. Harland & Wilson, 1956; Termier & Termier, 1956; Downloaded from https:/www.cambridge.org/core. Open University Library, on 17 Jan 2017 at 19:37:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0016756807003391 634 W. B. HARLAND Harland & Rudwick, 1964; Kirschvink, 1992; Hoffman Norwegian horizons (Harland, Hambrey & Waddams, et al. 1998). Hicks (1880) went on to consider the 1993; Harland, 1997). relationship between glaciation and volcanism. The best-preserved examples are in the Polaris- breen Group (Eastern Terrane) in the Hecla Hoek Complex. The upper tillite, Wilsonbreen Formation, 2.b. Norway correlates with Kulling’sSveanor tillite and Mortensnes In 1891 H. Reusch described a glacial moraine at Formation, of Finnmark. It is 200+ m below early Bigganjargga in Finnmark, north Norway, and referred Cambrian fossiliferous strata. A late Neoproterozoic to an ice age (istiden), then ?Cambrian (Strachan, age is consistent with acritarchs there (Knoll, 1981) 1897), now late Neoproterozoic. This small outcrop and in the Western Terrane (Knoll, 1992). Wilson & is now thought to be a slumped unit in a sandstone– Harland (1964) emphasized the conspicuous reddish quartzite formation with soft sediment sliding marks matrix of the tillite as evidence of erosion into a and boulder impressions (i.e. not a glaciated pave- lateritic basement, but Fairchild & Hambrey (1984) ment as he suggested). Other sliding lineations were found evidence that the reddening was secondary. observed at other levels in the formation (Harland, However these authors reinforced the earlier notion 1964b). This allochthonous tillite is no less indicative that profound climatic changes were represented at the of contemporary glaciation. Following a description of upper and lower boundaries of each of two glaciations, the Finnmark region by Holtedahl (1919), Føyn (1937) by the characteristics of the bounding dolomitic and Strand (1937), Bjørlykke (1967) established two carbonate strata. Together these observations reinforce distinctive tillite horizons: the lower (Smalfjord)˚ and the concept of a major Varanger glacial period in which upper (Mortensnes) formations. The lower unit was freezing temperatures interrupted dominantly warm correlated with Reusch’s moraine. They are separated environments (Harland, 1964a, b). Moreover, attempts by the reddish siltstone Nyborg Formation, all in the at palaeoclimatic maps as by Bain (1960) as well Vestertana Group (Vidal & Bylund, 1981). This tillite as more recent late Neoproterozoic–early Cambrian sequence was correlated with Sparagmite diamictites reconstructions (e.g. Unrug, 1997; Dalziel, 1997) place of southern Norway (Moelv Formation, Holtedahl Greenland on or near the palaeoequator. It was thus (1922)). no great surprise to obtain an early palaeomagnetic result for Greenland of 8◦ N (Bidgood & Harland, 1961; Harland, 1964a, b), although now known to be 2.c. East Greenland overprints (Evans, 2000). However, it was also clear in In East Greenland Poulsen (1930) and Kulling (1930) the 1960s that continental drift reconstructions could described the Tillite Group (Koch, 1929) with the Cape not accommodate all Neoproterozoic glacials as polar. Oswald Formation. They correlated the succession with In the Central Terrane of Spitsbergen a distinct two glacial horizons with those in Finnmark. This has stromatolite unit is interbedded with the later Varanger become a classic succession for research and detailed tillite. Stromatolites were regarded as exclusively correlation with Svalbard (Hambrey & Spencer, 1987; belonging to tropical carbonates, but Fairchild et al. Fairchild & Hambrey, 1995). (1989) showed that examples in the Wilsonbreen Formation resembled those in Antarctic lakes, recalling the earlier concept (Harland & Herod, 1975) of 2.d. Svalbard carbonate precipitation in extreme cold environments. A massive boulder unit was described in southwestern The Western Terrane, with two much thicker early Spitsbergen by Garwood & Gregory (1898). Kulling and late Varanger tillite formations, was then near (1934) described the Sveanor tillite northeast of Spits- to Ellesmere Island and adjacent to Trettin’s Pearya bergen (Nordaustlandet). He correlated it with East Terrane (1987) where Troelsen had identified tillite Greenland and Finnmark glacials and surveyed current (1950) and where these overall successions are well knowledge of late Precambrian tillites worldwide, matched (Harland, 1997). which is discussed further later. Reflecting on this and on other Scandinavian occurrences Kulling (1951) 2.e. Australia proposed a major Varanger Ice Age based on the Finnmark Varangerfjord succession. In South Australia H. P. Woodward in 1884 suspected The Svalbard tillites