Petrochemical Study of the Tephra Sequence Exposed in the Phira Quarry, Thera

INTRODUCTION

The tephra produced in the Bronze Age eruption of Thera is the uppermost of three pumiceous tephra deposits exposed in the Santorini island group, which have been designated the Lower, Middle, and Upper Pumice by Pichler and Kussmahl (1972). This Upper Pumice, up to 60 m thick according to Günther and Pichler (1973), consists of a distinct basal layer of pinkish lump pumice, the so-called Rose Pumice, and an upper part consisting of pumiceous ash which contains lumps of pumice up to 15 cm in diameter and fragments of older rocks, mostly in the higher levels.

The Rose Pumice, up to 5 m thick (Günther and Pichler 1973) is of pumice fall origin, judging by the lack of evidence of horizontally directed force and deposition on Mount Profitis Elias, the highest point on the island of Thera (Pichler  & Kussmahl 1972).

The tephra overlying the Rose Pumice is divisible into a lower bedded ash up to 7 m thick and an overlying chaotic ash which may be over 50 m thick (Günther & Pichler 1973). In the quarries at Phira and Oia and other exposures it is seen to be stratified and low-angle cross-bedded. Originally the bedded tephra was thought to have been produced by numerous mild intermittent explosive events separated by periods of erosion and redeposition, over an unknown period of time (Reck 1936). In the opinion of the volcanologists who visited Thera on the occasion of the First International Scientific Congress on the Volcano of Thera, the stratification and cross-bedding could be the result of base surge as described by Moore (1967), and that view has been adopted in recent works on the subject (Günther & Pichler 1973; Bond & Sparks 1976). The bedding, dune, and antidune features thus would not represent any break in continuity of the eruption.

The base surge layer grades upward into the chaotic tephra with an imperceptible or barely perceptible break in deposition. This chaotic layer has been interpreted in different ways: as an ash flow (Pichler & Kussmahl 1972; Günther and Pichler 1973), and as a mud flow (Bond & Sparks 1976).

In order to throw additional light on the nature of the Bronze Age eruption, and in particular to ascertain whether there are any significant discontinuities in composition of the tephra sequence, a series of samples collected from different levels in the Phira pumice quarry (Fig. 1) was studied. This paper presents the preliminary results of chemical, mineralogical and petrographic study of those samples. The levels in the quarry section from which the samples were collected are indicated in Fig. 2.

PETROGRAPHY

The samples all consist mainly of glass shards, with phenoclasts of feldspar, pyroxene, magnetite, and rarely, amphibole. Samples SQ - 1, SQ - 10, and SQ - 15 are of particular interest, being slightly higher in crystal content than the rest.

MINERALOGY

• Glass

The glass shards are of uniform shape, occasionally crescentic, clear, and contain very little opaque material. The range of refractive indices of the glass as determined by oil immersion is shown in Figure 3.

• Phenoclasts

The phenoclasts are predominantly plagioclase feldspar, clinopyroxene (augite), and orthopyroxene (hypersthene), with an alkali feldspar and magnetite in lesser amounts and amphibole in very minor amount. The abundance of each of the minerals (Fig. 4) was obtained by point count (200 points on each of the numbered samples).

• Plagioclase feldspar (An 30)

This is the commonest mineral, ranging from 4 to 7 percent of the samples. Although it usually occurs as individual crystals 0.5 - 1 mm long (with glass attached, in the samples from the ash), the plagioclase occasionally occurs in glomeroporphyritic groups. Optical zoning is not uncommon, and the cores of zoned crystals commonly show resorbed edges. Resorption is also indicated by embayed plagioclase boundaries as seen in thin-section.

• Alkali feldspar

A mineral having the properties of high-temperature albite (2V = 30°, optically negative, n = 1.530, inclined extinction) was encountered occasionally. Sometimes it displayed Carlsbad twinning. Fluctuations in the abundance curve for this high-temperature albite (?) (Fig. 4) have little significance due to the small percentage of the mineral observed in the point counts.

• Pyroxenes

The two pyroxenes present in the tephra, augite and hypersthene, together constitute 3 - 7 percent of the samples (see Fig. 4). Augite is dominant except at the top of the stratigraphic section (samples SQ - 1 & SQ - 2). The ratio of plagioclase to pyroxene content is roughly 2.5 : 1, both in the more crystal-rich levels (2.5 - 2.8 : 1) and in the others (2.5 : 1).

• Magnetite

Magnetite is not always common; occasionally it accounts for approximately 2.0 - 3.0 percent of the sample. It occurs mainly as tiny grains scattered through the glassy part of the pumice.

• Amphibole

Amphibole is encountered on rare occasions (not more than one crystal per sample). The rarity suggests that it may be of accidental origin, incorporated into the samples by stripping from the throat of the vent.

PETROCHEMISTRY

• Major elements:

Twenty-three new analyses for major oxides, determined by wet-silicate analysis, are given in Table 1. Twenty-one of the analyses represent samples from the stratigraphic collection of the Bronze Age tephra from the Phira quarry (SQ - 1 is from the top to SQ - 48 from the base); the last two (A -  1 and A - 8) are samples of lump pumice collected from the Akrotiri excavations and from their vicinity.

Certain oxides, specifically SiO₂, K₂O, CaO, TiO₂ and P₂O₂ show distinct variations from the top to the bottom of the section. MgO and Na₂O also vary, but to a lesser extent.

The variation in SiO₂ (63.7 - 67.3 percent) is the most pronounced of any of the oxides (Fig. 5). The SiO₂ values are higher at the base and lower at the top of the section. The variation of the other oxides with respect to SiO₂ is shown in Fig. 6. 

In order to determine the relative influence of SiO₂ content on the refractive index, a portion of each of 18 samples was fused in a carbon arc according to the method of Mathews (1951), and the resulting beads crushed for refractive index determination. Figure 7 shows the refractive indices of the fused beads plotted against SiO₂ content; the curve of Mount Lassen samples derived in a similar fashion by Mathews (1951) is shown for comparison. In Figure 8 the refractive indices of the fused beads are plotted against the stratigraphic level of the sample. The silica-rich (basal) samples from the lower part of the stratigraphic section cluster at the opposite end of the trend line from the silica-poor samples collected from the upper part of the section. Note that the beads of the two crystal-rich levels (SQ - 15 and SQ - 10) plot near the silica-poor levels from the top of the stratigraphic section. 

• Alkali Oxides

Alkali variation is due almost entirely to change in K₂O content, which decreases slightly toward the upper part of the stratigraphic section and increases with increasing silica content (see Fig. 5).

• Ferric Oxides

The total iron content in the Bronze Age Tephra increases from the base to the top of the section (see Fig. 5). However, breaks occur in the pattern (e.g. SQ - 15, SQ - 10 the crystal-rich samples). The plot of total iron vs. SiO₂ (see Fig. 9) indicates a distinct trend of iron depletion in the more silica-rich levels. The iron-poor, silica-rich levels are in the lower parts of the section, and locally in layers containing smaller modal percentages of crystals (SQ - 12, SQ - 13). Samples at the top of the section (SQ - 1, SQ - 2), and those with larger modal percentages of crystals (SQ - 15) fall in the iron-rich end (see Fig. 7).

The MgO content (.80 to 1.73 percent) also increases toward the top of the section (see Fig. 5). Minor increases also occur in the crystal-rich levels (SQ - 15, SQ - 10). MgO content decreases as silica increases.

•  Other Major Oxides

The second most prominent change noted in the Bronze Age Tephra is that of CaO (2.8 - 3. 46 percent) (see Fig. 5); the increase in this instance probably is due to the increase in modal plagioclase (see Fig. 4). Modal feldspar percentages in samples from the top of the section are higher than those at the midsection; the modal pyroxene content, on the other hand, does not vary.

Variations in other oxides are generally of a minor nature, but in general they are in accord with the changes recorded in the major oxides. Al₂O₃ and TiO₂ are examples (see Table I).

Two other variation diagrams, A-F-M (Fig. 10) and K-Ca-Na (Fig. 11), confirm a progressive change in the Bronze Age tephra.

Within the 37 m exposed in the Phira quarry, the spread of the plotted points of the tephra along the calc-alkaline trend in the expanded A-F-M diagram is noticeable. The glass-rich levels, and samples from the base of the section as well, plot in the alkalic region and samples from the top of the section as well as from the crystal-rich layers plot in the more Fe-Mg rich region. In addition, slight iron enrichment with decreasing total alkalis is indicated.

The expanded K-Ca-Na diagram also shows a noticeable degree of differentiation within the Bronze Age tephra. Samples from near the base of the section as well as samples relatively rich in glass plot in the alkalic part of the trend, and the crystal-rich portions of the section plot at the calcium-rich end.

TRACE ELEMENT DATA

X-ray fluorescence data were obtained for four trace elements, Sr, Zr, Rb, and Y. These data are plotted against stratigraphic level in Fig. 12. Zr and Y show no essential difference in the top and bottom samples; Sr and Rb contents decrease from the base to the top of the section and are probably related to the modal plagioclase content.

DISCUSSION

Chemical, optical, and mineralogical data on samples of the Bronze Age tephra from a stratigraphic section in the Phira quarry show a regular decrease in SiO₂ and KO₂ content and an increase in MgO, CaO, TiO₂, P₂O₅ and total iron content from the base to the top of the section, and a variation of the content of modal glass and minerals which suggests a partial explanation for the observed distribution of the major oxides.

Taken together, these results are consistent with the concept of magma chamber inversion (during the eruption the magma chamber was emptied from the top down; thus the upper part of the stratigraphic section in the Phira quarry would represent the magma from the lower levels of the chamber and the lower part of the section, the higher levels). A similar pattern of behavior in the chemistry and mineralogy of tephra deposits, attributed to magma chamber inversion, has been described from Nevada by Lipman and others (1966).

The greatest chemical variations do not occur at the boundaries between the three lithologic units of the Bronze Age tephra (Rose Pumice, base-surge layers, and chaotic tephra). Rather, they reflect small changes in the concentration of crystals of plagioclase and pyroxene against a background of an otherwise gradual change in chemical composition from rhyodacitic to dacitic as the eruption progressed.

The interruption of the general trend by the concentration of crystals represented by the levels SQ - 10 and SQ - 15 suggests that simple crystal settling in a fractionating melt may not have been the only process to have operated in the case of the Thera eruption. These variations may be attributed to crystal showers resulting from the tapping of localized concentrations of crystals in the magma. The refractive index of the glass from the levels in question is essentially unchanged from that of the glass on the adjacent levels, indicating that the composition of the melt proper did not change substantially. The variations observed for these levels, therefore, must be the product of mechanical localization of crystals.

The absence of any significant change in chemistry or mineralogy that would correspond to the change from the Rose Pumice to the base surge deposits suggests that there was no interruption of activity, merely a change in style of the eruption. The lack of any change whatsoever in chemistry and mineralogy that could mark the change from the base-surge deposits to the overlying chaotic tephra, together with the lack of a definite stratigraphic break, argues even more forcefully for the origin of the latter as an ash flow, the two together representing the same (presumably climactic) paroxysmal phase of the eruption.

If the chaotic tephra were the product of mud flows, a more pronounced stratigraphic break should be visible at its base.

One further word concerning the possibility that the Rose Pumice could represent a separate eruption occurring some 50 years before the rest of the tephra was ejected. Smith (1960) and Smith & Bailey (1966) have discussed the problem of the filling of a magma chamber and have concluded that the time required for the refilling and re-eruption of a chamber is of the order of at least 10,000 years. The lack of any evidence for weathering or incipient soil development anywhere within the section exposed in the quarry further supports the idea that the eruption was a single continuous event.

-----------------------------------

--------------------------------------