The origin of tektites - settled at last?
News & Views from Peter J. Smith, Nature Vol. 300 18 November 1982
Tektites, those small glassy objects limited to a few well-defined regions of the Earth's surface, have haunted geologists for more than a century. Like country house ghosts, they have never been part of the investigative mainstream but have given rise to a variety of weird and wonderful explanations that attract the occasional attention of the wider community particularly when it has seemed that a new technique might throw some light on their nature and origin. Unlike country house ghosts, on the other hand, they are undoubtedly real, having been regarded as fine ornamental material since at least the time of Cro-Magnon man, about 31,000 years ago.
It was perhaps unfortunate that early scientific attention should have been directed towards Czechoslovakian tektites (moldavites) from a region of ancient European glassmaking. The result, during the nineteenth and early twentieth centuries, was a long time-wasting debate over whether tektites are real or artificial, productive only in the sense that the right conclusion was reached in the end. By then, however, it had already become clear that if tektites were natural they were not to be easily explained, for there was no obvious source, or 'mother lode'. The way was thus open to a wide range of possible explanations from the down-to-earth (literally and figuratively) to the bizarre.
Terrestrial volcanism was the obvious first choice; but it soon became evident that there were significant chemical differences between tektites and obsidian, the more common glassy material of undisputedly volcanic origin. Besides, candidate volcanoes close to the tektite zones, or 'strewnfields', were hard to find. To avoid some of these difficulties, Dunn (Victoria, Geol. Surv. Bull. 27, 3; 1912) proposed an elaborate derivation of tektites from obsidian, involving lightning strikes in a very dusty atmosphere packed with hot gases, but it was not sufficient to overcome the objection that tektite compositions were far removed from those of terrestrial volcanics generally. During subsequent decades, tektites were to be attributed to, among other things, the heating of stony meteorites, the collision between a meteorite and a natural Earth satellite, the drying-out of siliceous gels acted upon by humic acid, and the fall of antimatter, although neither these nor many other hypotheses stood the test of time, if they ever stood at all.
What we are left with in 1982 are just two contending hypotheses, both of which involve extraterrestrial bodies. According to one, tektites are volcanic products ejected from the Moon; and according to the other, they are fragments ejected from the Earth itself by cometary or meteoritic impact. As work on returned lunar samples has proceeded, a lunar volcanic origin of tektites has come to seem less and less likely, although there are those, most notably O'Keefe (Geochim. cosmochim. Acta 44, 2151; 1981), who would strongly dispute that view. The alternative explanation, by contrast, has risen in popularity, bolstered no doubt by a natural desire to avoid exotic solutions wherever possible. The hot money these days is on the idea that tektites result from the impact melting of terrestrial rocks, expecially sediments; and the odds in favour of that must surely have risen with the publication by Shaw and Wasserburg (Earth planet. Sci. Lett. 60, 155; 1982) of a detailed study of the world's tektites using Sm-Nd and Rb-Sr systematics.
There are four tektite strewnfields, within each of which the tektites have the same K -Ar and fission track ages. Thus the North American, European, Ivory Coast and Australasian tektites are about 35, 14, 1.3 and 0. 7 million years old respectively. These ages represent the times at which the tektites were last molten; they say little about the types of rock from which the tektites were formed, how they formed, or whence they came. Fortunately, however, it is now possible to determine rather more, both comparatively and absolutely, about the provenance and evolution of rocks from the behaviour of the Sm-Nd and RbSr decay systems.
As for the comparative, one way of expressing isotopic ratios is to use the e notation. Thus eNd (0) is the measured deviation (in parts in 104) of the I43Nd/144Nd ratio from the present-day chondritic value of0.511847; and e5' (0) is the deviation (also in parts in 101) of the 87Sr/86 Sr ratio from a model undifferentiated mantle reservoir value of 0.7045. The point about e values is simply that they are different for different rock types. Shaw and Wasserburg have measured e values for geographically and geochemically representative samples from all four tektite strewnfields and find that eNd (0) is always negative and e5' (0) is always positive and large. These values are characteristic of old terrestrial continental crust. They exclude continental material which has been derived from the mantle very recently, for that has e values close to zero; and they exclude oceanic crust, because that has positive eNd (0) values and negative esr (0) values. The isotopic data thus match the chemical data, for, as many workers have pointed out since at least 1917, the chemistry of tektites resembles nothing so much as continental sediments, especially sandstones. There are no known lunar rocks having major and trace element compositions similar to those of tektites, and there are no known meteorites with these compositions.
The second important discovery made by Shaw and Wasserburg is that within each tektite group eNJ (0) is uniform whereas f:s' (0) is highly variable. This reflects what earth scientists already suppose about the different types of information provided by the Sm-Nd and Rb-Sr systems. Assuming that concentrations are high enough to measure, any radioactive decay system will give an age for the rock concerned, but the term 'age' needs careful interpretation. The major fractionation of Sm and Nd appears to occur during the partial melting of mantle material to produce crust, and the Sm-Nd system is relatively undisturbed by such processes as weathering and sedimentation. Moreover, Sm and Nd are refractory and thus not readily fractionated by differential volatilization on impact melting. The Sm-Nd age of a tektite should therefore represent the time of formation of the crustal material from which the tektite was ultimately derived, and uniformity of eNJ (0) within a single tektite group inspires confidence that the Sm-Nd system is dating precisely that event. The Rb-Sr system, by contrast, can be severely disturbed by sedimentation, volatilization, weathering and diagenesis. It is only to be expected, therefore, that a tektite group will provide a range of f: 5' (0) values and Rb-Sr ages reflecting the varied history of the tektite source rocks and descending to the last major parentdaughter fractionation event. In short, if nature is behaving itself, all tektites within a group should have a common Sm-Nd age, a range of younger Rb-Sr ages and, of course, a common and even younger K-Ar age representing the time at which the tektite themselves formed.
And thus it turns out. The Sm-Nd ages obtained by Shaw and Wasserburg for the North American, European, Ivory Coast and Australasian tektites are, respectively, about 650, 900, 1,900 and 1,150 million years. In other words, all the ages are consistent with the tektites' having been formed from Precambrian crust. The RbSr data then indicate what happened between the formation of that crust and its conversion to tektite. The European tektites, for example, were derived from sediments laid down only about 20 million years ago. Interestingly, sediments in the nearby Ries Crater have E values and isotopic compositions similar to those of the tektites themselves. Moreover, the age of the oldest basement in the crater area agrees well with the Sm-Nd age of the tektites. The new data thus support an earlier suggestion that the Ries Crater was formed by the impacting body that also gave rise to the tektites.
The Ivory Coast strewnfield has also been associated with an impact structure, namely, the Bosumtwi Crater. The rocks around this crater have ages comparable with the Sm-Nd age of the tektites. The Rb-Sr systematics indicate that the tektites were formed from Precambrian sediments about 950 million years old, although the sediments have not been identified yet. The North American tektites, by contrast, have no known source crater. Moreover, the fact that the material that went to form the tektites was derived from the mantle as recently as the late Precambrian rules out most of the North American Precambrian shield, and the sediments derived therefrom, as a source region . Similar age criteria refute the recent suggestion from Dietz (Meteoritics 12, 145; 1977) that the Popigai astrobleme of Siberia is the source of the North American strewnfield. The only known continental material of the required age is the late Precambrian crust constituting the east-southeast margin of the Appalachian orogenic belt, although no suitable crater has been identified within it.
Little more can be said about the origin of the North American tektites, except that the immediate source material was sedimentary, for the Rb-Sr proved unsuitable for the determination of an age of sedimentation. Not the least interesting conclusion to be drawn from the work of Shaw and Wasserburg, however, is that an oceanic impact for the origin of tektites is not ruled out by the restriction that tektite source material cannot be oceanic crust. Although the effects of the impact at the Ries Crater extend to depths of about 2 km, the tektites supposedly produced by the impact were formed from sediments which appear to have been only a few tens of metres thick at the time. The implication of this is that an oceanic impact could well generate tektites simply by sampling thin continentally derived sediments lying on top of the basaltic oceanic crust. In other words, an impact origin of tektitites is not precluded by failure to discover a suitable impact structure on land.
In the case of the Australasian strewnfield there are several candidate craters on land - the Zhamanshin and Elgygytgyn Craters of the USSR and an unnamed crater-like structure in Cambodia. Nothing much is know of the Cambodian crater and the Elgygytgyn Crater can be ruled out on age grounds. The most touted possibility is the Zhamanshin Crater, but the case for it is weak, if not nonexistent. For example, this crater is associated with tektite-like objects known as irghizites, which have chemical compositions similar to, but Nd isotopic compositions quite different from, those of the tektites of the Australasian strewnfield. It is possible in principle that both the irghzites and the Australasian tektites were produced at the same crater but that the irghizites managed to take up a proportion of the impacting meteoritic material, although Shaw and Waserburg show that it is impossible in practice to obtain the correct e values for the irghizites simply by adding chondritic material to the tektitic. It seems much more likely that the impact body that supposedly gave rise to the Australasian tektites fell in the ocean; but wherever it struck, it generated tektites from continentally derived sediment about 250 million years old.
What is impressive about the story pieced together by Shaw and Wasserburg is its overall coherence and internal consistency. So can we take it that the tektite issue is now settled? Or will the lunar volcanologists fight back, as they have so often done before, with a quite different interpretation of the same data? We shall see. In the meantime, it could well be objected that, coherent or not, the cometary impact hypothesis is no less exotic a mechanism for the production of tektites than is lunar volcanism. But somehow it does seem less so these days. Recent calculations on the frequency with which comets large and small might be expected to reach the Earth's surface, and the remarkably cool way in which the earth science community has been willing to consider seriously the suggestion by Alvarez eta/. (Science208, 1095; 1980)that cometary impact may have been responsible for the Cretaceous-Tertiary extinctions, suggest that the rigid anticatastrophism characteristic of geological thinking since the time of Hutton and Lyell is now dead.
Yet at least one puzzle remains, and remains unremarked. If cometary impacts have been responsible for biological extinctions throughout the Phanerozoic, why have they apparently only generated tektites during the past 35 million years? Curious.
It was perhaps unfortunate that early scientific attention should have been directed towards Czechoslovakian tektites (moldavites) from a region of ancient European glassmaking. The result, during the nineteenth and early twentieth centuries, was a long time-wasting debate over whether tektites are real or artificial, productive only in the sense that the right conclusion was reached in the end. By then, however, it had already become clear that if tektites were natural they were not to be easily explained, for there was no obvious source, or 'mother lode'. The way was thus open to a wide range of possible explanations from the down-to-earth (literally and figuratively) to the bizarre.
Terrestrial volcanism was the obvious first choice; but it soon became evident that there were significant chemical differences between tektites and obsidian, the more common glassy material of undisputedly volcanic origin. Besides, candidate volcanoes close to the tektite zones, or 'strewnfields', were hard to find. To avoid some of these difficulties, Dunn (Victoria, Geol. Surv. Bull. 27, 3; 1912) proposed an elaborate derivation of tektites from obsidian, involving lightning strikes in a very dusty atmosphere packed with hot gases, but it was not sufficient to overcome the objection that tektite compositions were far removed from those of terrestrial volcanics generally. During subsequent decades, tektites were to be attributed to, among other things, the heating of stony meteorites, the collision between a meteorite and a natural Earth satellite, the drying-out of siliceous gels acted upon by humic acid, and the fall of antimatter, although neither these nor many other hypotheses stood the test of time, if they ever stood at all.
What we are left with in 1982 are just two contending hypotheses, both of which involve extraterrestrial bodies. According to one, tektites are volcanic products ejected from the Moon; and according to the other, they are fragments ejected from the Earth itself by cometary or meteoritic impact. As work on returned lunar samples has proceeded, a lunar volcanic origin of tektites has come to seem less and less likely, although there are those, most notably O'Keefe (Geochim. cosmochim. Acta 44, 2151; 1981), who would strongly dispute that view. The alternative explanation, by contrast, has risen in popularity, bolstered no doubt by a natural desire to avoid exotic solutions wherever possible. The hot money these days is on the idea that tektites result from the impact melting of terrestrial rocks, expecially sediments; and the odds in favour of that must surely have risen with the publication by Shaw and Wasserburg (Earth planet. Sci. Lett. 60, 155; 1982) of a detailed study of the world's tektites using Sm-Nd and Rb-Sr systematics.
There are four tektite strewnfields, within each of which the tektites have the same K -Ar and fission track ages. Thus the North American, European, Ivory Coast and Australasian tektites are about 35, 14, 1.3 and 0. 7 million years old respectively. These ages represent the times at which the tektites were last molten; they say little about the types of rock from which the tektites were formed, how they formed, or whence they came. Fortunately, however, it is now possible to determine rather more, both comparatively and absolutely, about the provenance and evolution of rocks from the behaviour of the Sm-Nd and RbSr decay systems.
As for the comparative, one way of expressing isotopic ratios is to use the e notation. Thus eNd (0) is the measured deviation (in parts in 104) of the I43Nd/144Nd ratio from the present-day chondritic value of0.511847; and e5' (0) is the deviation (also in parts in 101) of the 87Sr/86 Sr ratio from a model undifferentiated mantle reservoir value of 0.7045. The point about e values is simply that they are different for different rock types. Shaw and Wasserburg have measured e values for geographically and geochemically representative samples from all four tektite strewnfields and find that eNd (0) is always negative and e5' (0) is always positive and large. These values are characteristic of old terrestrial continental crust. They exclude continental material which has been derived from the mantle very recently, for that has e values close to zero; and they exclude oceanic crust, because that has positive eNd (0) values and negative esr (0) values. The isotopic data thus match the chemical data, for, as many workers have pointed out since at least 1917, the chemistry of tektites resembles nothing so much as continental sediments, especially sandstones. There are no known lunar rocks having major and trace element compositions similar to those of tektites, and there are no known meteorites with these compositions.
The second important discovery made by Shaw and Wasserburg is that within each tektite group eNJ (0) is uniform whereas f:s' (0) is highly variable. This reflects what earth scientists already suppose about the different types of information provided by the Sm-Nd and Rb-Sr systems. Assuming that concentrations are high enough to measure, any radioactive decay system will give an age for the rock concerned, but the term 'age' needs careful interpretation. The major fractionation of Sm and Nd appears to occur during the partial melting of mantle material to produce crust, and the Sm-Nd system is relatively undisturbed by such processes as weathering and sedimentation. Moreover, Sm and Nd are refractory and thus not readily fractionated by differential volatilization on impact melting. The Sm-Nd age of a tektite should therefore represent the time of formation of the crustal material from which the tektite was ultimately derived, and uniformity of eNJ (0) within a single tektite group inspires confidence that the Sm-Nd system is dating precisely that event. The Rb-Sr system, by contrast, can be severely disturbed by sedimentation, volatilization, weathering and diagenesis. It is only to be expected, therefore, that a tektite group will provide a range of f: 5' (0) values and Rb-Sr ages reflecting the varied history of the tektite source rocks and descending to the last major parentdaughter fractionation event. In short, if nature is behaving itself, all tektites within a group should have a common Sm-Nd age, a range of younger Rb-Sr ages and, of course, a common and even younger K-Ar age representing the time at which the tektite themselves formed.
And thus it turns out. The Sm-Nd ages obtained by Shaw and Wasserburg for the North American, European, Ivory Coast and Australasian tektites are, respectively, about 650, 900, 1,900 and 1,150 million years. In other words, all the ages are consistent with the tektites' having been formed from Precambrian crust. The RbSr data then indicate what happened between the formation of that crust and its conversion to tektite. The European tektites, for example, were derived from sediments laid down only about 20 million years ago. Interestingly, sediments in the nearby Ries Crater have E values and isotopic compositions similar to those of the tektites themselves. Moreover, the age of the oldest basement in the crater area agrees well with the Sm-Nd age of the tektites. The new data thus support an earlier suggestion that the Ries Crater was formed by the impacting body that also gave rise to the tektites.
The Ivory Coast strewnfield has also been associated with an impact structure, namely, the Bosumtwi Crater. The rocks around this crater have ages comparable with the Sm-Nd age of the tektites. The Rb-Sr systematics indicate that the tektites were formed from Precambrian sediments about 950 million years old, although the sediments have not been identified yet. The North American tektites, by contrast, have no known source crater. Moreover, the fact that the material that went to form the tektites was derived from the mantle as recently as the late Precambrian rules out most of the North American Precambrian shield, and the sediments derived therefrom, as a source region . Similar age criteria refute the recent suggestion from Dietz (Meteoritics 12, 145; 1977) that the Popigai astrobleme of Siberia is the source of the North American strewnfield. The only known continental material of the required age is the late Precambrian crust constituting the east-southeast margin of the Appalachian orogenic belt, although no suitable crater has been identified within it.
Little more can be said about the origin of the North American tektites, except that the immediate source material was sedimentary, for the Rb-Sr proved unsuitable for the determination of an age of sedimentation. Not the least interesting conclusion to be drawn from the work of Shaw and Wasserburg, however, is that an oceanic impact for the origin of tektites is not ruled out by the restriction that tektite source material cannot be oceanic crust. Although the effects of the impact at the Ries Crater extend to depths of about 2 km, the tektites supposedly produced by the impact were formed from sediments which appear to have been only a few tens of metres thick at the time. The implication of this is that an oceanic impact could well generate tektites simply by sampling thin continentally derived sediments lying on top of the basaltic oceanic crust. In other words, an impact origin of tektitites is not precluded by failure to discover a suitable impact structure on land.
In the case of the Australasian strewnfield there are several candidate craters on land - the Zhamanshin and Elgygytgyn Craters of the USSR and an unnamed crater-like structure in Cambodia. Nothing much is know of the Cambodian crater and the Elgygytgyn Crater can be ruled out on age grounds. The most touted possibility is the Zhamanshin Crater, but the case for it is weak, if not nonexistent. For example, this crater is associated with tektite-like objects known as irghizites, which have chemical compositions similar to, but Nd isotopic compositions quite different from, those of the tektites of the Australasian strewnfield. It is possible in principle that both the irghzites and the Australasian tektites were produced at the same crater but that the irghizites managed to take up a proportion of the impacting meteoritic material, although Shaw and Waserburg show that it is impossible in practice to obtain the correct e values for the irghizites simply by adding chondritic material to the tektitic. It seems much more likely that the impact body that supposedly gave rise to the Australasian tektites fell in the ocean; but wherever it struck, it generated tektites from continentally derived sediment about 250 million years old.
What is impressive about the story pieced together by Shaw and Wasserburg is its overall coherence and internal consistency. So can we take it that the tektite issue is now settled? Or will the lunar volcanologists fight back, as they have so often done before, with a quite different interpretation of the same data? We shall see. In the meantime, it could well be objected that, coherent or not, the cometary impact hypothesis is no less exotic a mechanism for the production of tektites than is lunar volcanism. But somehow it does seem less so these days. Recent calculations on the frequency with which comets large and small might be expected to reach the Earth's surface, and the remarkably cool way in which the earth science community has been willing to consider seriously the suggestion by Alvarez eta/. (Science208, 1095; 1980)that cometary impact may have been responsible for the Cretaceous-Tertiary extinctions, suggest that the rigid anticatastrophism characteristic of geological thinking since the time of Hutton and Lyell is now dead.
Yet at least one puzzle remains, and remains unremarked. If cometary impacts have been responsible for biological extinctions throughout the Phanerozoic, why have they apparently only generated tektites during the past 35 million years? Curious.