With the exception of rock shelters, the caves of Maryland show a tendency to develop along joints. Faults, cleavages, or fractures other than joints are apparently not significant in controlling the direction of passages in solution-type caves. In most caves one set of joints exerts major control over the pattern with the larger passages developed along them, and subordinate side passages follow the secondary joints.
      In flat-lying limestones in Garrett County, caves are simple in pattern. Generally, one main passage is developed that follows a major set of joints with occasional offsets along subordinate joints. Multiple levels are confined to local sections of caves and are connected by vertical cliffs or shafts. Passages slope uniformly but show no consistent relation to the dip of the rocks.

Cave developed at joint intersection

      In folded strata, where caves lie on the flanks of folds, passages develop as fissure-like openings along steeply inclined joint planes. The passages vary from a few feet to over 100 feet high and consist of several parallel openings. Except in Twiggs Cave, the passages are offset along the dip where they occur in more than one level. In Twiggs Cave the joint control is so dominant that levels are not offset but are developed one above the other.
      Caves lying near the crests of anticlines, like Revells Cave and Crystal Grottoes, have a maze of interlacing passages equally developed along two or more sets of joints. This results in a plan resembling city blocks.
      With the exception of Sand Cave and Devils Den, bedding exerts little control in cavern development in Maryland except to modify the cross-sectional shape of some passages. In Sand Cave the bedding determines the extent of the cave, and joints tend to modify the shape of the walls, but are otherwise unimportant in controlling the pattern. In Devils Den the bedding determines the direction of the major passages.
      It has been observed that a predominance of Maryland caves exhibit a marked horizontal nature and extent. Very few have more than 100 feet of vertical relief. Many have sloping passages but for the most part remain on one plane. The control of pattern by the strike of the rocks is not constant. At high dip, however, strike-oriented patterns are common but by no means the rule. In flatter limestones, there is no preferred pattern (e.g. Crabtree Cave).
      These views are further substantiated by Barr (1961), who classified Tennessee caves into two major types according to inclination of the beds. He proposed the terms Allegheny and Appalachian to designate these types. Allegheny type caves are developed in strata of relatively low dip (e.g. Garrett County). The floors and ceilings tend to be horizontal except where secondarily modified by fill, breakdown, and domepit. Major passages tend to follow the dip, with patterns often being complex and different levels superimposed.
      Appalachian type caves appear in steeply or moderately inclined beds and are the predominant pattern in the Appalachian Valley (Washington and Allegany Counties). The floors and ceiling of different levels are not superimposed, as a rule, but tend to occur down-dip from, and usually parallel to, other levels. Devils Hole Cave in Allegany County is a good example of this.
      In summary it may be said that the caves of the Appalachian Ridge and Valley province are typically single conduit type passages parallel to the ridges, with large lateral extents being rare. The limestones do not have protective cap rocks, so the caves are subject to relatively easy truncation and collapse processes. The limestones of the interior plateaus, in contrast, are nearly flat-lying and occur often under sandstone or shale-capped ridges. These caves are related to much larger drainage basins, often by as much as a factor of 10 (Poulson and White, 1969), than those caves in the Appalachians and have a relatively larger lateral extent.

(from ES3 -Caves of Maryland)