figure 2 
OBSERVATIONS ON SILICIFICATION IN GEODES FROM THE MISSISSIPPIAN SANDERS GROUP OF INDIANA AND KENTUCKY
Henry L. Barwood and Nelson R. Shaffer
Indiana Geological Survey
Bloomington, Indiana 47405
ABSTRACT
Geodes are common in certain units of the Mississippian Sanders Group in Indiana and Kentucky. They are most abundant in silty dolomitic units that have the highest contents of acid (HCl) insoluble residues of quartz, clay minerals, and feldspar. Petrographic examination of geodes shows concentric zones of quartz that grade from microcrystalline and fibrous varieties at the edge to coarse mosaic crystals that rim interior mineralization. The interior macrocrystalline quartz varieties usually contain gypsum or anhydrite inclusions. Partially silicified anhydrite nodules (proto-geodes) from a newly created exposure of Borden and Ramp Creek equivalents near Louisville, Kentucky, show that silicification occurs contemporaneously with precipitation of anhydrite. The outer edge of these proto-geodes preserve banded microcrystalline quartz that grades into spherulitic fibrous quartz. Interior to the microcrystalline border are coarse quartz crystals that have zoned cores with anhydrite inclusions and rims that are inclusion-free. This macrocrystalline quartz frequently has a pseudocubic habit and usually is coated with a thin layer of length slow chalcedony that grades into zebraic chalcedony. We speculate that anhydrite nodules precipitated at or near the sediment water interface in response to calcium saturation brought on by a combination of dolomitization and sulfate-reduction of magnesium sulfate-rich brines. The sulfate reduction generated the ubiquitous ferrous sulfides and supplied ferrous iron to glauconite and ferroan smectites. An increase in pH accompanying the sulfate reduction phase probably mobilized opaline silica from sponge spicules. This silica source is consistent with oxygen isotope data (-24 to -29 o/oo vs. SMOW) that suggests that the silica was from the same sources as the insoluble residues. Silica solubilization lagged behind anhydrite precipitation leaving the anhydrite nodules with sulfate-rich cores. Later removal of these sulfate cores generated voids that are a distinguishing characteristic of geodes and created sites for subsequent precipitation of a host of minerals.
WIDESPREAD DISTRIBUTION OF GEODES IN MISSISSIPPIAN LIMESTONES
Long before their recognition as silicified anhydrite nodules, geodes were recognized from Mississippian limestones in Iowa, Missouri, Illinois, Indiana, Kentucky, Tennessee, Alabama and Georgia. Formations generally equivalent to the Sanders Group are known to contain geodes where they outcrop throughout the southeast and mid-west. Petrographic examination of the geodes eventually revealed the traces of anhydrite that provided clues as to their origin (Chowns and Elkins, 1974). Mineralization in geodes is extremely varied and ranges from simple quartz linings to unusual minerals to solid chalcedony fillings (Milliken,1979, Dott and Batten, 1971). Areas away from basins tend to produce geodes that are solid and filled with colorful banded chalcedony known as agate (Zeitner, 1964). Silicified anhydrite type geodes have also been reported from a number of other locations worldwide, especially in the Mesozoic (Tucker, 1976, Prichystal, et al. 1999)
DISTRIBUTION OF GEODES IN THE SANDERS GROUP
The Sanders Group consists of the Ramp Creek Formation , the Harrodsburg Limestone and the Salem Limestone. The Ramp Creek is an interbedded dolosiltite (in places a dolomitic mudstone) and limestone with abundant geodes and chert (Nicoll and Rexroad, 1975). The Harrodsburg Limestone is dominantly bedded limestone with intervals of shale and argillaceous limestone (Carbonate Petrology Seminar, 1987). Geodes are not as abundant within the Harrodsburg. The Salem Limestone is a high energy limestone formed from a carbonate shoal environment and geodes are very rare in the Salem (Brown, 1990). Geodes are abundant in fine grained dolomitic facies of the Ramp Creek and Harrodsburg and much less abundant in coarser grained high energy environments (Maliva, 1987).
The Sanders Group represents a variety of facies and the while the Salem Limestone is pervasive across the Illinois Basin, the Ramp Creek and Harrodsburg grade laterally into the Ft. Payne and Ullin Formations to the west and into the Muldraugh and Ft. Payne Formations to the South (Lasemi., et al., 1998). Overall the Sanders Group is best represented by a carbonate ramp model (Ahr, 1973, Saturni, 1985). Actual conditions in the Sanders Group probably represent a combination of a ramp that eventually drops off the Borden Delta into deep water and a shelf with a Sabhka environment, separated by a shoal complex.
The geode-dolomite association and the distribution of geodes by facies was noted even before the anhydrite nodule origin was discovered (Hayes, 1964). A key assumption in previous investigations was that the anhydrite nodule precursors to the geodes formed from reflux after burial and precipitated from brines generated by the overlying St. Louis Limestone. This model assumed that the anhydrite nodules formed after sediment deposition but prior to lithification and generally assumes that the nodules were displacing the surrounding sediments during growth. There is actually little evidence for displacement and most nodules show growth on or near the sediment surface (figure 1,2). (Click on the thumbnails to view full size images. All images are crossed polars unless otherwise noted)
figure 1 figure 2
MINERALOGICAL CHARACTERISTICS OF THE SILICIFICATION EVENT
Comparison of silica mineralization from the Overlying St. Louis formation with similar silica mineralization in the Sanders Group reveals some interesting differences. Chert and silicified fossils from the St. Louis show a high concentration of fine fibrous quartz spherulites and replaced limestone (figure 3,4,5). These spherulites contain mostly length slow chalcedony and apparently originated from evaporitic conditions (Folk and Pittman, 1971). Many other quartz forms have been reported from silica associated with evaporite conditions (Siedlecka, 1972, Keene, 1983). What is notable about the silica from the St.Louis Formation is the absence of anhydrite inclusions in the megaquartz
figure 3 
figure 4 
figure 5
The earliest silica mineralization in the geodes is also spherulitic, but records coarse spherulites considered to be replacements of anhydrite (figure 6) and banded spherulites that suggest crystallization from a gel state (figure 7,8). Rarely, primary gypsum crystals are embedded in this spherulitic geode rim (figure 9). The spherulites contain abundant anhydrite inclusions and grade into anhedral quartz that also contains anhydrite inclusions (figure 10,11,12). Occasionally, a spherulite develops into a radiating anhedral megaquartz group (figure 13,14). This initial silicification probably releases significant dissolved anhydrite and evidence of this release can be found in veins of fibrous gypsum that are often located at the top of dolosilt and mudstone beds (figure 15).
figure 6 
figure 7
figure 8 
figure 9
figure 10
figure 11
figure 12 
figure 13
figure 14
figure 15
Some of the protogeodes found along highway 313 near Fort Knox, Kentucky contain relict anhydrite that shows several textures. Interlocking coarse fibers of anhydrite are most common (figure 16) but some examples have curved fibers radiating from quartz crystals (figure 17). These radiating anhydrite fibers may have developed in response to a chemical or pressure gradient during quartz growth. Another interesting feature observed in the protogeodes is organic material that accumulated during quartz growth (figure 18,19).
figure 16 
figure 17
figure 18
figure 19
The defining characteristic of the quartz found in the geodes and protogeodes is anhydrite inclusions. Typically anhydrite is scattered throughout the megaquartz, but often it is concentrated in the cores of the crystals (figure 20,21). Two other silica forms are also characteristic of geodes, so called
ACubic@ quartz and twisted fiber (Zebraic) chalcedony. The Cubic@ quartz is characteristic of evaporite deposits and is frequently reported from evaporites (Tarr and Lonsdale, 1929). While this type of quartz can develop from simple rhombohedral faces, there is evidence that the geode Acubes@ are twinned crystals consisting of Brazil law twins (figure 22,23).figure 20
figure 21
figure 22 
figure 23
Twisted fiber (Zebraic) chalcedony is also characteristic of geodes and often encrusts the quartz crystals of the cavities in layers up to several centimeters thick. The twisted fiber chalcedony is frequently separated from the
ACubic@ quartz by a thin layer of length slow chalcedony and development of this layer has been observed in the protogeodes (figure 24). The fibers initially align themselves with a crystallographic axis of the megaquartz (figure 25). As they grow, the fibers twist about the fiber axis and this generates the unique banded pattern observed in thin sections (figure 26).figure 24
figure 25
figure 26
SOURCE OF THE SILICA
Previous investigators have reported Oxygen isotope data in the range of -24 to -32.2 o/oo vs. SMOW for various silica components in geodes (Milliken, 1979). These values have been interpreted as reflecting deposition of the quartz forms in the range of 30 to 40 degrees centigrade. Data reported for the twisted fiber silica (Zebraic chalcedony) reflected a different temperature and isotopic composition and was considered an indicator of late deposition of this phase. This investigation studied the oxygen isotopic composition of the geodes and insoluble residues from the surrounding limestones. The data showed a range of oxygen isotope compositions of -24 to -29 o/oo vs SMOW for both the geodes and the insoluble residues. This suggests that both the geodes and the insoluble component of the limestone shared a similar source, assumed to be sponge spicules.
MODEL FOR GEODE FORMATION SUGGESTED BY THE SILICIFICATION
Previous investigators have suggested that the anhydrite nodules that are precursors to silicified anhydrite nodule (geode) formation were deposited during dolomitization of the sediments (Maliva, 1987). This is in agreement with both field and laboratory observations of the protogeodes and geodes. The silica was introduced into the anhydrite nodules either during precipitation of the anhydrite or very shortly thereafter. The presence of gel banding in the earliest silica deposited indicates that the silica was present as silica gel and suggests a high pH environment. The geodes and protogeodes are also closely associated with pyrite and other sulfide minerals, organic matter and either glauconite or ferroan smectite (nontronite). This suggests an anoxic environment at the time of formation.
A model that fits all these characteristics would consist of a shelf evaporite environment that provides brines denser than sea water. These brines would develop in open oxygenated environments (the lower St. Louis Limestone) and need not be highly concentrated. The brines would sink and flow through the porous sediments of the shoal complex (Salem Limestone) and continue down dip through the Harrodsburg Limestone and Ramp Creek until they reached the edge of the deeper anoxic portion of the Illinois Basin. Here they would
Apuddle@ and become anoxic. Dolomitization would commence with the finer grained sediments in the mud mounds and silty calcareous mudstones driving the concentration of calcium to the point where anhydrite would begin to precipitate. Contemporaneously with this dolomitization there would be an increase in pH from anoxia and sulfate reduction. This would generate sulfide minerals, reduced clays and take opaline silica into solution. The silica would be available for replacement of the anhydrite nodules and for silicification of limestone to form chert. Where silica concentrations were high, silica gel probably precipitated, as is observed in some geode rims.The anoxic zone would sweep across the ramp in response to changes in the depth of the anoxic zone and, combined with facies influenced dolomitization, would generate irregular concentrations of geodes, as observed on outcrop. In the deeper portions of the basin mostly bedded cherts developed as in the Ft. Payne Formation. The more oxygenated, higher energy formations such as the upper part of the Harrodsburg and the Salem Limestones rarely contain either geodes or dolomite.
REFERENCES
Ahr, W.M. 1973. The Carbonate Ramp: An Alternative to the Shelf Model. Transactions-Gulf Coast Association of Geological Societies 23:221-225
Brown, M.A. 1990. Regional Lithofacies and Depositional Environments of the Salem Limestone (Valmeyeran, Mississippian), South-Central Indiana. In: Thompson, T.A. (Editor) Architectural Elements and Paleoecology of Carbonate Shoal and Intershoal Deposits in the Salem Limestone (Mississippian) in South-Central Indiana. Indiana Geological Survey Guidebook 14:1-12
Carbonate Petrology Seminar. 1987. Ramp Creek and Harrodsburg Limestones: A Shoaling-Upward Sequence with Storm-Produced Features in Southern Indiana, U.S.A. Sedimentary Geology 52:207-226
Chowns, T.M. and J.E. Elkins. 1974. The Origin of Quartz Geodes and Cauliflower Cherts Through the Silicification of Anhydrite Nodules. Journal of Sedimentary Petrology 44(3):885-903
Fisher, I.S. 1977. Distribution of Mississippian Geodes and Geodal Minerals in Kentucky. Economic Geology 72:864-869
Folk, R.L. and J.S. Pittman. 1971. Length-Slow Chalcedony: A New Testament for Vanished Evaporites. Journal of Sedimentary Petrology 41(4):1045-1058
Hayes, J.B. 1964. Geodes and Concretions from the Mississippian Warsaw Formation, Keokuk Region, Iowa, Illinois, Missouri. Journal of Sedimentary Petrology 34(1):123-133
Keene, J.B. 1983. Chalcedonic Quartz and Occurrence of Quartzine (Length-Slow Chalcedony) in Pelagic Sediments. Sedimentology 30:449-454
Lasemi, Z., R.D. Norby and J.D. Treworgy. 1998. Depositional Facies and Sequence Stratigraphy of the Lower Carboniferous Bryozoan-Crinoidal Carbonate Ramp in the Illinois Basin, Mid-Continent, U.S.A. In: Wright, V.P. and T.P. Burchette (editors) Carbonate Ramps. Geological Society, London Special Publication 149:369-395
Maliva, R. G. 1987. Quartz Geodes: Early Diagenetic Silicified Anhydrite Nodules Related to Dolomitization. Journal of Sedimentary Petrology 57(6):1054-1059
Milliken, K.L. 1979. The Silicified Anhydrite Syndrome- Two Aspects of Silicification History of Former Evaporite Nodules from Southern Kentucky and Northern Tennessee. Journal of Sedimentary Petrology 49(1):245-256
Nicoll, R.S. and C.B. Rexroad. 1975. Stratigraphy and Conodont Paleontology of the Sanders Group (Mississippian) in Indiana and Adjacent Kentucky. Indiana Geological Survey Bulletin 51. 36 p.
Prichystal, A, Z. Losos and D. Richterova. 1999. Genesis of Jurassic Silica Geodes in Moravia. European Journal of Mineralogy 11:183
Saturni, B.A. 1985. Depositional History of the Middle Valmeyeran Ramp Creek and Harrodsburg Limestones of Southwest Indiana. Unpublished M.S. Thesis. Indiana University, Bloomington, Indiana. 78 p.
Siedlecka, A. 1972. Length-Slow Chalcedony and Relicts of Sulphates- Evidences of Evaporitic Environments in the Upper Carboniferous and Permian Beds of Bear Island, Svalbard. Journal of Sedimentary Petrology 42(4):812-816
Tarr, W.A. and J.T. Lonsdale. 1929. Pseudo-Cubic Quartz Crystals from Artesia, New Mexico. American Mineralogist 14:19-25
Tucker, M.E. 1976. Quartz Replaced Anhydrite Nodules (
ABristol Diamonds@) from the Triassic of the Bristol District. Geological Magazine 113(6):569-574Zeitner, J.C. 1964. Midwest Gem Trails. Gembooks, Mentone, CA 80 p.