Conclusions

1) The sills of the Central Arizona Sill Complex are flow-differentiated, particularly with respect to olivine but also with respect to orthopyroxene in a single very important case, without exception among those from which completely sampled sections exist. While plagioclase phenocrysts may also exhibit flow differentiation (lower Roosevelt Dam sill), between the middle of a sill, where the mafic phenocrysts or xenocrysts are most numerous, and the base, may be found a relatively narrow layer of concentrated plagioclase phenocrysts the origin and existence of which is presently difficult to explain. The best present interpretation is that this layer results from crystallization front fractionation (e.g. Marsh, 1992).

2) The lower unit at Pocket Creek, the Sierra Ancha sill complex of Nehru (1961) and Nehru and Prinz (1970), consists of at least three separate intrusions instead of the two or possibly three identified by those authors, and the contacts are differently located. These intrusions include a layer at the top of the unit which is approximately 100 ft. (30.5 m) thick, the centrally located microtroctolite zone, and the balance of the sill.

3) The existence of a thick, previously unknown, sill located above the lower unit at Pocket Creek is reported. It was not possible to obtain a lower chilled margin from this sill, but it is geochemically associated with the Tholeiitic suite. With this, and the second conclusion, above, the finding of Fouts (1974) that only the sills at Theodore Roosevelt Dam and in the Salt River Canyon were formed from multiple intrusions is not sustained.

4) The sill at Theodore Roosevelt Dam, like that of the lower unit at Pocket Creek, is divided into three cooling units instead of the two recognized by Fouts (1974).

5) Contrary to previously published opinion (Smith, 1969; Nehru and Prinz, 1970; Dostal and Fratta, 1977), the chilled margins of sills in the Central Arizona Sill Complex do in fact represent liquid compositions. While it is likely that K2O, Rb, Ba, and possibly Th have been modified by interaction with continental material, according to extant data for K2O and Rb this has certainly not happened at the site of emplacement.

6) Because of the low MgO, Ni, and Cr contents of the rocks, none of them can be considered to represent primitive liquids.

7) Based on geochemistry, three separate suites have been recognized within the Central Arizona Sill Complex: The Tholeiitic suite, Alkaline suite #1, and Alkaline suite #2. The Tholeiitic suite appears to correlate with the high-SiO2 suite of Fouts (1974), while the alkaline suites correlate with his low-SiO2 suite.

8) The spatial separation of the low-SiO2 and high-SiO2 suites reported by Fouts (1974), to the north and south of the Globe-Miami area, respectively, is not confirmed. The lower unit at Pocket Creek belongs to Alkaline suite #2 and the upper unit to the Tholeiitic suite. Other high-SiO2 rocks occur as far north as the Salt River Canyon. It is true, however, that the Alkaline suite rocks, as sampled, are entirely to the north of this index.

9) The members of the Tholeiitic suite can be related to one another by fractional crystallization. The more extensive variation among the members of Alkaline suite #1 cannot be accounted for either by fractional crystallization or magma mixing, and it may be that its members are the result of separate, distinct, igneous events, that there has been selective assimilation of crustal material, or that they have been affected by some other process. There is little variation among the members of Alkaline suite #2 and none which can be attributed to igneous processes; for this reason, it is suggested that Alkaline suite #2 may represent a single intrusive event. The suggestion of Fouts (1974) that the Tholeiitic suite (high-SiO2) was derived from the same parental liquid as the Alkaline suites (low-SiO2) is not confirmed.

10) The possibility cannot presently be ruled out that there has been contamination by continental crustal material arising from a mantle source which has been modified by a subduction event; indeed, the ubiquitous negative Nb anomaly in the absence of contamination of the magma by known or reasonable crustal compositions suggests that such may be the case at least for the Alkaline suites. Apropos this possibility, Kempton et al. (1990) argue on the basis of stable and unstable isotopic evidence that a sedimentary or other crustal component exists in lower crustal meta-igneous, probably subduction-related, rocks of the region which are probably Proterozoic in age. The contamination effect is much more apparent in the Tholeiitic suite than in either of the Alkaline suites, and it may be that the Tholeiitic suite interacted directly with continental crust at some point in its history. Shastri and Bowring (1991), working with isotopic data from diabasic pegmatites of the Sierra Ancha and post-orogenic dikes from northwest and central Arizona, have indicated that assimilation of older crust is very much a possibility; but compositional data from their samples do not exist and therefore geochemical affiliation is unknown. From the intrusive habit, it may be that they are Tholeiites. The same contamination effects could be produced, however, if the Tholeiitic suite, and to a much lesser extent the Alkaline suites, underwent contamination of the primary magma prior to the latest fractionation event.

11) From a plot of Ti/Y versus La/Nb, it may be that the three identified suites originate from different regions of the mantle having distinctive geochemical characteristics. It is not possible reasonably to generate the most evolved suite (the Tholeiitic suite) from the most primitive (Alkaline suite #2) by fractional crystallization or AFC, nor is it possible reasonably to derive Alkaline suite #1 from Alkaline suite #2 by this means. Alkaline suite #1, however, lies on a mixing line between the other two in this space. It is therefore possible that there are two distinct melting regions involved, represented by the Tholeiitic suite and Alkaline suite #2, and that a combination of their characteristics gave rise to Alkaline suite #1. Clearly, more work in this direction is needed before a definitive statement can be made.

12) It is speculated that the progressive positive P anomaly noted in Alkaline suite #1, opposite in sense to that of the Tholeiitic suite, may be the result of some mantle enrichment process such as mantle metasomatism. This may also explain the failure of attempts to model compositional variations within Alkaline suite #1. In any of the Ti/Y versus La/Nb diagrams (e.g. Figures 128 and 131), it is evident that the direction of increasing "evolution" of Alkaline suite #1 runs counter to those of AFC and fractional crystallization trends. Mantle metasomatism of the source might account for this, and is also consistent with the observation that degree of partial melting interpreted from Figures 160 through 162 decreases with increasing "evolution." (For the present, it is necessary to refer to the full dissertation for copies of the figures cited.)

13) The pressures of the latest major fractionation events are approximately 8 kb for Alkaline suite #1, very approximately 8 kb for Alkaline suite #2, and near-surface for the Tholeiitic suite. The 8 kb events may correspond to the base of the ancient crust. Because of the likelihood that clinopyroxene was fractionated (low Cr content, and cpx component required in the fractional crystallization models relating members of the suite), it is probable that the Tholeiitic suite fractionated to some unknown and variable extent at higher pressure also. Fouts (1974) reached very similar conclusions in that he felt his low-SiO2 rocks had differentiated at a depth of 30 km (roughly 10 kb) and that the high-SiO2 rocks did so at much shallower levels.

14) Based on geochemistry, the rocks of the Central Arizona Sill Complex are most closely related to continental flood basalts.

15) The close similarity which exists between rocks of the Central Arizona Sill Complex and those from known continental rifting environments supports the suggestion of Burke and Dewey (1973) that these, and sills of similar composition and age which exist between Arizona and the Yukon in similar host rocks, occupy the failed third arms of RRR triple junctions which coalesced in an ancient rifting event in what is now the western United States. This conclusion is reached notwithstanding the suggestion of Howard (1991), based on structural theory, that the region was undergoing compression or was at least in an isotropic condition of stress at the time of intrusion of the diabase. The possibility is recognized that the similarity cited could exist because many of the rocks used in comparison were emplaced through Precambrian crust, and therefore are associated with "old" lithosphere.

16) Though their relationship with the balance of the Tholeiitic suite cannot presently be determined, two members of that group were intruded into two of the more primitive members of Alkaline suite #1 and are therefore younger than their hosts.

17) Based on a search of the literature, the ages of all of the members of the Central Arizona Sill Complex are not a settled issue. Published ages of 1.1 billion years (e.g. Silver, 1960, 1963) can be associated with only one of the three suites identified in the CASC, Alkaline suite #2. In this context, recall the work of Creasey (1965) in the vicinity of San Manuel, Arizona, in which he identified two generations of diabasic activity which he felt were Precambrian in age and one which he thought might be Cretaceous. Helsley and Spall (1972) show the co-incidence of paleomagnetic poles determined at various sites between Globe and the north rim of Salt River Canyon, which suggests similarity of ages; but again, it is impossible to relate these results to particular suites identified here. It may eventually be found that the sills of the CASC do in fact have the same or closely similar ages, but this is not established by extant data.

18) The distance measure (I) of Sneath and Sokol (1973; see also Brower and Zar, 1977) is a useful method for comparing rock compositions. It has the advantage of being dimensionless and therefore can be used to compare the closeness of agreement of entire analyses, including both major and trace elements. The relative lack of sensitivity imposed by the constraint that the index vary between zero and one could be overcome by multiplying by a constant, say 1000. In its present form, an I-value of 0.040 appears to be equivalent to a sum of the squares of the residuals for the major elements of approximately 1.

19) Probably the most important contribution of this work is the re-establishment of the CASC as a viable field of study: Previous investigators (e.g. Smith, 1969; Nehru and Prinz, 1970; Dostal and Fratta, 1977) had believed the sills which comprise it to be contaminated by host rocks, and therefore inscrutable. It is true that there are serious gaps in the data set used here; but this is perhaps a strength, not a weakness, if the work is recognized for what it is: A foundation for future efforts. Additional trace-element data are needed (Ta, Hf, more and better quality REE analyses, etc.); serious, comprehensive isotopic work (Pb, Rb, Sr, K, Nd, Sm, Os, etc.) has not yet begun; more crystal chemical data and better sample coverage are needed within individual sills. There is almost never enough information, but even the longest journey begins with a single step. These are indicators of the next which should be taken to improve the present understanding of the CASC, a continuing effort in which this work represents only my beginning.