Samples were chosen for analysis following initial transmitted light petrographic examination of thin sections, followed by more detailed examination of polished thin sections cut after sampling of core.

Despite examination of over 100 thin sections covering virtually all Leg 38 samples, Cr-rich spinel was found only in plagioclase-phyric lavas from a small portion of Leg 38 Hole 338. A general petrographic description of these rocks is given in Talwani, Udintsev, et al. (1976), but, because of sampling problems, these lavas are described in greater detail. Numerous errors exist in core photograph captions given throughout Talwani, Udintsev, et al. (1976). For the spinel-bearing cores, the published photograph of Section 38-338-43-2 is actually Section 38-338-42-3; similarly, the photograph of Section 38-338-43-3 is actually 38-338-43-2, and that of Section 38-338-43-4 is actually 38-338-43-3. The true photograph of Section 38-338-43-4 wasn't published. Cr-spinel is found within two lava flows within a mapped three-lava-flow sequence within these sections. This sequence was originally mapped as a single dike or sill by Talwani, Udintsev, et al. (1976). The top of the sequence is characterized by a nearly totally altered breccia zone; the three moderately to highly altered lava flows (here designated as Units 1, 2, and 3) are differentiated by vesicle-rich zones in the manner described for the East Greenland SDRS lavas (Larsen, Duncan, Allan et al., 1995). Like the East Greenland SDRS lavas, they erupted subaerially judging from presence of the breccia zone and the lack of marginal quench. Cr-rich spinel from Leg 38 samples are found exclusively as occasional, subhedral to euhedral inclusions, from 20 to 100 µm in size, within Unit 1 and 2 plagioclase phenocrysts. Examples of the shapes and sizes of these inclusions are given in Figure 2. Unit 1, extending from Section 38-338-42-2, 66 cm to Section 38-338-43-4, 110 cm, consists of moderately to highly plagioclase phyric basalt with a fine-grained, highly altered, holocrystalline groundmass. Plagioclase phenocrysts make up 5%-10% of the rock, are up to 3 mm in size, and commonly are found as glomerocrysts up to 1 cm in size. Isolated crystals of anhedral-subhedral clinopyroxene and euhedral olivine are greatly subordinate, making up <1% of the rock and are <1 mm in size. Olivine and the fine-grained mesostasis (30%-35%) are totally altered to greenish-brown clays. Unit 2, extending to Section 38-338-45-2, 34 cm, is less crystal rich and much less glomeroporphyritic, with 2%-3% blocky plagioclase phenocrysts (up to 5 mm) and larger amounts of anhedral-subhedral clinopyroxene and euhedral olivine (2%-3% each, up to 2 mm in size). Olivine and the fine-grained mesostasis (15%-20%) are totally altered to greenish-brown clays. Unit 3, extending to Section 38-338-45-2, 150 cm, does not contain Cr-rich spinel.

Cr-rich spinel was reported from a relatively primitive Leg 104 dike sample and as "traces of chromite as inclusions" in other unnamed samples (Viereck et al., 1989). Nevertheless, subsequent examination by the first author at the ODP East Coast Core Repository of this dike sample and of 86 other Leg 104 thin sections covering the entire recovery found no evidence of Cr-rich spinel. A few samples from the relatively primitive chemical formation USC in Hole 642E Upper Series (with whole-rock Mg#s that average 0.68 and Cr that may exceed 300 ppm; Viereck et al., 1989) should have been able to precipitate Cr-rich spinel (e.g., Allan et al., 1988), but apparently did not. Most Leg 104 lavas are too evolved to carry Cr-rich spinel, especially those of the evolved Lower Series (Parson et al., 1989).

Leg 81 thin sections were unavailable for examination at the time of this study, but no reports of finding Cr-rich spinel in these samples have been made in the literature (Harrison et al., 1984; Joron et al., 1984; and Richardson et al., 1984). Moreover, few Leg 81 samples reported in these studies have Mg# higher than 0.60 and Cr higher than 300 ppm (most are well below 200 ppm Cr), with these more primitive samples being of a doleritic character and, therefore, unlikely to carry Cr-rich spinel (because of Cr partitioning into clinopyroxene and making the Cr-rich spinel unstable).

Cr-rich spinel is absent from the Leg 163 lavas and from sites other than Site 917, cored during Leg 152. In contrast, petrographic analysis of the more primitive, olivine-rich lavas of Hole 917A shows them to have abundant Cr-rich spinel. Because of the extensive descriptions of these samples in Larsen, Saunders, Clift, et al. (1994), only a short, slightly modified summary is presented here. It should be noted that the Leg 152 shipboard party described unpolished thin sections, leading to inevitable underestimates of the occurrence and abundance of Cr-rich spinel.

Larsen, Saunders, Clift, et al. (1994), amplified by Fitton et al. (1998a, 1998b), Larsen and Saunders (1988), and Saunders et al. (1998), divided the igneous recovery at Hole 917A into 92 volcanic units grouped into three subaerial igneous series. The Lower Series, comprised primarily of pahoehoe and aa basaltic lavas, and the Middle Series, comprised of evolved aa lavas and several ashflows, erupted from large, long-lived, crustally contaminated reservoirs, with shallowing of the crustal reservoirs with time. The generally more primitive Upper Series consists of thinner pahoehoe flows showing little evidence of crustal contamination and were likely erupted from smaller, more ephemeral reservoirs. Larsen, Saunders, Clift, et al. (1994) speculated that the Upper Series was derived from partial melting of more MORB-like mantle, and it represented the final phase of continental breakup of the southeast Greenland margin, with Fitton et al. (1998b) emphasizing that most of the Hole 917A suite was derived from MORB-like mantle. Other work (Fram et al., 1998) proposes that the Upper Series is actually more complex and can be broken into two groups derived from normal MORB and depleted Icelandic plume mantle, respectively.

Cr-rich spinel is found in nearly all of the aphyric olivine basalt, olivine-phyric basalt, and picrites that make up the Upper Series. Olivine phenocrysts are the differentiating feature of these lavas, with picrites defined as containing >15% olivine phenocrysts in hand sample (or >25% olivine phenocrysts in thin section) and showing evidence of olivine accumulation (Larsen, Saunders, Clift, et al. 1994). Olivine-phyric basalts were defined as containing from 1%-15% olivine phenocrysts in hand specimen (15%-22% olivine in thin section); olivine phenocrysts are uncommon but visible in hand specimen in the most commonly present aphyric olivine basalt (but contain 5%-18% olivine in thin section).

Cr-rich spinel is absent from the evolved Middle Series and from the more evolved aphyric basalts and plagioclase-dominated basalts of the Lower Series. Nevertheless, it is found abundantly in the more primitive picrites and olivine-phyric basalts of the Lower Series, and is also found in the Lower Series aphyric olivine basalts, olivine-plagioclase-phyric basalts (1%-5% olivine phenocrysts and 1%-8% plagioclase phenocrysts), and olivine-plagioclase-pyroxene basalts (>1% olivine, plagioclase, and clinopyroxene phenocrysts).

The groundmass of both Upper and Lower Series lavas may be quite coarse, with subophitic and holocrystalline, intergranular textures commonly defined by the groundmass plagioclase, clinopyroxene, and Ti-rich magnetite. They are moderately to heavily altered, with all of the olivine and fine-grained mesostasis altered to secondary clays, with plagioclase and pyroxene (including occasional orthopyroxene) usually being partially altered as well. The most olivine-rich samples-the picrites and olivine-phyric basalts-are typically the most altered.

Cr-rich spinel is most abundantly found as euhedral inclusions in olivine or as euhedral crystals embedded along the sides or edges of olivine, but loose groundmass crystals are also common. Back-scattered electron (BSE) images of representative Cr-rich spinel crystals are given in Figure 3 and Figure 4. Cr-rich spinel is most abundant and largest in size in the most primitive samples (e.g., the picrites and olivine-phyric basalts). Groundmass Cr-rich spinels (Fig. 3) are jacketed by thick rims of chromian magnetite (ferritchromite) and Ti-rich magnetite, with the Ti-rich magnetite often exhibiting trellis exsolution of ilmenite (Sample 152-917A-11R-4, 45-50 cm, in Fig. 3). These spinels commonly show dissolution textures in conjunction with the chromian magnetite and Ti-rich magnetite jackets (Fig. 3B, C), with near-total degradation of smaller Cr-rich spinel grains (Fig. 3B). In some cases, these textures could possibly represent rapid Ti-rich magnetite growth and formation of chromian magnetite in concert with Cr-rich spinel dissolution (P. Roeder, pers. comm., 1996). Ti-rich magnetite often nearly completely replaces the previous Cr-rich spinel, evidenced by only very small relict cores; as a result, it is difficult to avoid underestimation of Cr-rich spinel magmatic occurrence. Cr-rich spinel is best preserved in the Hole 917A lavas when it is wholly enclosed within olivine, as crystal surfaces embedded within olivine lack Ti-rich magnetite rims.