Opisthosomal appendages and gill lamellae
The results herein are in agreement with earlier histological observations that the book gills arise as epithelial evaginations from the posterior surface of the branchial appendages [11, 32]. Kingsley  also suggested there is some 'intucking' that might become more pronounced in producing the internal and anterior-directed air sacs of arachnid book lungs. In the present study there was no indication of epithelial invagination in the formation of individual gill lamella. Those authors [11, 32] did not identify their tissue sections with a specific embryonic stage, since the stages 1-21 were characterized in later research [25, 26].
From small bilateral ridges in the ventral surface of the opisthosoma (Figure 1A) , the distal part of the genital operculum and branchial appendages extend ventrally and become bilateral, immobile flaps in the third embryonic molt (Figure 1B) [1, 32]. This clearly increases the surface area for gas exchange, and more striking changes continue in the interval (stage 20) between the third and fourth embryonic molts. The cellular outgrowth at the tip and along the length of the distal part of these appendages results in their knobby appearance at this stage (Figures 2A, B and 3, 4).
After the third embryonic molt (stage 20), a narrow cleft is evident between the bases of the opisthosomal appendages (Figures 2B, 3) as their proximal regions lengthen from the ventral surface of the opisthosoma. A thicker and distinctive proximal part of the appendages is formed. This basal region is filled with a diversity of cells and does not initially have a central lumen (Figures 3, 5, 8, 9). Some cells are deteriorating as though in the process of forming a lumen (Figures 3, 8). Other cells may migrate toward the periphery to become part of the hypodermal layer.
The gill lamellae begin as small cell clusters or an outward fold of hypodermis and cuticle at the posterior surface of the thick basal region of the first branchial appendage (Figures 1B, 5, 6, 7, 9, 12B) . The development of the gill lamellae appears to be an elaboration of the general pattern of evaginating epithelial cell growth that is evident in the knobby appearance of the opisthosomal appendages in their early stages of formation (Figures 2A, B and 3, 4).
The basic structure of the gill lamellae is essentially the same as for the genital operculum and branchial appendages (Figures 2A, B and 3, 5, 6, 7, 8, 9), as though these structures are the result of a common program from the opisthosomal appendage anlage. That program includes the following: l) evagination and outgrowth of epithelium (Figures 5, 6, 9), 2) formation of a sac-like structure with central lumen (hemolymph channel) and pillar-type trabeculae (Figures 2A, B and 7, 9, 10, 11, 12A, B), 3) a hypodermal layer of epithelial cells (Figures 3, 7) and 4) typical apical-basal polarity of these epithelial cells, i.e., apical secretion of cuticle and the basal surface in contact with the hemolymph (Figures 3, 7, 14) [1, 2, 42, 43].
While the gill lamellae and distal parts of the opisthosomal appendages are ectodermal and epithelial (Figures 2A, B and 7, 9, 10, 11), the thick basal part of the opisthosomal appendages has mesodermal derivatives: hemocytes (Figures 6, 8), muscle (Figures 9, 12A) and endoskeleton (Figures 12A, B and 13). In their diagrams of embryo histological sections, Patten and Hazen  show that each opisthosomal segment has bilateral mesosomal somites, and the ventral cells of the somite are the source of the slender shaft of endoskeleton that eventually extends into the basal part of the genital operculum (Figures 12A, 13) and first branchial appendage (Figure 12B).
The endoskeleton has a thin electron-opaque matrix, and typically many matrix chambers have cell debris or cells with large vacuoles (Figure 13). This suggests the endoskeleton is deteriorating, but vacuolated cells are a common feature of invertebrate cartilage [37–39] including the cartilage of adult horseshoe crabs .
From observations herein, some endoskeleton is a common feature of the basal part of the genital operculum, present even in the first and second instars that were examined. Patten and Hazen  explained that their endoskeletal descriptions are based mainly on the genital operculum, and they assumed a similar pattern also occurs in the branchial appendages. They also note that the endoskeleton of the first branchial appendage is later in development than the endoskeleton of the genital operculum.
After the third and fourth embryonic molts in the present study, some endoskeleton is present at the base of the first branchial appendage in some specimens (Figure 12B), but in others there are large vacuoles that appear to be the remains of deteriorating cells (Figures 9, 12A). The endoskeleton may not yet have been formed in these latter examples, or possibly there is variable development of endoskeleton in these early stages.
From immobile flap-like structures after the third embryonic molt (Figures 1B, 2A, B), the appendages with gill lamellae become highly mobile after the fourth molt (Figures 7, 9, 12B) [1, 32]. The somewhat variable endoskeletal conditions observed herein suggest there are changes in the endoskeleton associated with the onset of appendage mobility. This would occur especially in the endoskeleton of the genital operculum and first branchial appendage since these appendages have much development before their mobility begins after the fourth embryonic molt (stage 21).
Since the endoskeleton is produced from cells of the ventral somite , the proximal end of the endoskeletal shaft extends initially into the opisthosomal ventrum where the endoskeleton could impede appendage mobility. It is reasonable that some endoskeleton deteriorates or fails to form, especially at the proximal end (e.g., asterisk in Figure 12A), as the appendages become longer and mobile after the fourth embryonic molt. The longest lengths of endoskeleton were seen after the third embryonic molt (stage 20) while much smaller lengths were seen in later stages (Figure 12A, B). An endoskeleton has not yet been demonstrated at the base of the second and third branchial appendages that develop after the embryonic molts and hatching.
More detailed studies are needed to clarify the changing structure and role of the cartilage-like endoskeleton in these early stages. The evolutionary relationship of vertebrate and invertebrate cartilage is presently receiving much research attention ; also to be considered is the cartilage-like endoskeleton in the embryos and early instars of these 'living fossil' xiphosurids [44, 45].
Endoskeletal tissue regarded as different types of cartilage is present in numerous invertebrates and at the base of opisthosomal appendages in adult horseshoe crabs [31, 32, 35–39]. The branchial endoskeletal matrix in adult horseshoe crabs has elastin, allowing support but also flexibility for the rhythmically moving appendages with attached gills . The matrix that encloses the cells has protein fibrils in a metachromatic ground substance. The metachromatic staining property indicates the likely presence of the mucopolysaccharide, chondroitin sulfate . This is further supported in biochemical studies [38, 47, 48].
The pattern of cuticle formation evident in Figure 14 was commonly seen at the tip of outgrowing cell processes of the appendages and gill lamellae. Small, dense granules appear to be produced and released at the cell membrane, and these granules aggregate to form cuticle. Since the membrane particles differ in size from those inside the cell (Figure 14), there may be membrane enzymes that synthesize cuticle particles from precursor substances in the cell. A similar mode of cuticle formation is suggested in other arthropod developmental studies.
In an SEM investigation of book lung development in scorpion embryos, the cuticle of the air sac lamellae results from an aggregation of granular material released from the precursor cells . Similar SEM observations are reported for cuticle formation for book gill development in the horseshoe crabs . In a subsequent TEM study of scorpion book lungs, the initial cuticle of the air sac lamellae develops from particles formed at the plasma membrane of cell fragments released from the apical end of aligned precursor epithelial cells . For spermatophore development in male scorpions, the cuticle is formed by joining small osmiophilic plates produced at the tips of microvilli of adjacent epithelial cells . Some insect embryonic cuticle is apparently formed from small dense plaques produced by hypodermal cells with microvilli . At the biochemical level in insects, a membrane enzyme (chitin synthase) polymerizes cytoplasmic precursor molecules and transfers the resulting chitin to the extracellular space .
In post-third molt embryos, cuticular vesicles are commonly seen at the tips of the flap-like genital operculum and first branchial appendage (Figures 2A, 15). The thin, transparent membrane that holds the vesicles in place is probably exuvium still attached to these appendages after the third molt. The exuvium from the embryonic molts remains inside the external embryonic membrane [1, 26] and may be only partially removed from the opisthosomal appendages, especially in the early stages when appendages are not yet active.
No vesicles were seen at the tip of the second branchial appendage (Figure 2A, B), but this structure is relatively small in post-third molt embryos and may have been formed in the third molt (Figures 1B, 2A, B). After the third embryonic molt, numerous cuticular invaginations and evaginations are common at the tips of the genital operculum (asterisks in Figure 15) and first branchial appendage, as though these are sites of substantial growth. In Figure 15, the cuticle of the released vesicles is thinner than that of the nearby appendage tip, suggesting the cuticular vesicles are released earlier, i.e., during or just after the third embryonic molt. The significance and developmental role of these vesicles could not be determined from the light and electron micrographs examined herein.
The oldest fossil horseshoe crab is from the Lower Ordovician , and in their review of the fossil record, Rudkin and Young  ask the question why the body plan of the horseshoe crab has been so persistent while others have not. From observing first and second instars in the laboratory, it appears the book gills are easily clogged with debris and ill-suited for long-term survival of the body plan.
The cuticle of the opisthosomal appendages of first instars has channels that probably release a protective substance (Figures 16, 17). An exocuticle, layered endocuticle and pore canals are present in the cuticle of adult horseshoe crabs . The outer surface of their carapace is protected by an anti-fouling agent apparently secreted through pores in the cuticle . Specialized epithelial cells at the base of transcuticular pores are described as the likely source of a slimy covering especially present in younger horseshoe crabs [31, 54]. Gill cleaning behavior  and numerous moltings [26, 56] may also help prevent clogging and infection of the page-like gills. Leibovitz and Lewbart  provide an informative review of the infections of the gills and other parts of horseshoe crabs.
Scorpions are another group with a 'relict' body plan. The oldest fossil from the Silurian period [45, 58] looks much like modern forms. As reviewed by Dunlop , there has been much discussion about the phylogenetic relationship of scorpions and other chelicerates. Of course, the developmental sequence of animals is not a reliable indicator of their evolutionary history, but scorpion development should be considered since they have a conserved body plan. Pectines are commonly recognized as a distinctive feature of scorpions [58, 59]. The pectines begin to develop very early in the embryo, suggesting an early separation of a clade with these structures. The indicators of terrestrialization, spiracles and book lungs, appear much later in the embryo after the pectines are prominent .
The late appearance of the book lungs in scorpion embryos can be compared with the late appearance of gill lamellae in the developmental sequence of structures in the opisthosoma of horseshoe crabs. For the latter, the results herein show the following sequence: 1) bilateral ridges in each opisthosomal segment (Figure 1A), 2) thin distal lobes of the opisthosomal appendages (Figures 1B, 2A), 3) thick proximal base of the opisthosomal appendages (Figure 2B), 4) eventual formation of gill lamellae, muscle bands and endoskeleton (Figures 5, 6, 7, 9, 12A, B and 13)  and 5) rhythmic appendage movement with swimming and aeration of the gills. The ability to add increasing numbers of gill lamellae as the animal increases in size, and the ability to use the opisthosomal appendages for swimming and for passage of water over the gills have surely been major factors for their survival.
Comparison of development of book gills and scorpion book lungs
Muscle cells are not present within the scorpion book lungs, but there is some rhythmic agitation  produced by a process completely different from movement of book gills. Hypocardial ligaments extend from the ventral surface of the scorpion heart to the dorsal surface of the book lung sinus so that rhythmic heart contractions cause rhythmic movement of the book lungs. The hypocardial ligaments also contain muscle fibers which may be stretch-activated since no nerve fibers are evident. The book lungs, of course, have no endoskeleton.
In the formation of book gills, the primary result of precursor cell activity is the posterior-directed hemolymph sacs enclosed with cuticle and supported by pillar-type space holders (Figures 5, 6, 7, 9, 12A, B) . The water channels without space holders are a secondary result of the repeated parallel hemolymph sacs (Figures 9, 12B) . The sac-like hemolymph lamellae are formed mainly by proliferation and evagination of precursor epithelial cells followed by secretion of cuticle and development of a central hemolymph lumen (Figures 5, 6, 7, 9, 12B).
In the formation of scorpion book lungs, the primary result of precursor cell activity is the formation of internal, anterior-directed air sacs enclosed with cuticle and supported by specialized and diverse types of trabeculae [1, 2, 34, 49]. The air sacs are produced by inward migration (ingression) of precursor cells from the parent invaginated epithelium (atrial wall) rather than evagination from an appendage base as occurs for book gills [1, 2]. The hemolymph channels with pillar-type trabeculae are a secondary result of the parallel air sacs with their distinctive space holders.
For the development of scorpion air sacs, there is initial formation of parallel rows of precursor cells, and these cell rows anterior to the spiracle and atrium are the basis for the page-like air sacs [2, 49]. The aligned cells release (apocrine secretion) cell fragments from their apical surface. Since the precursor cells are in parallel rows, their secreted cell fragments are also in parallel rows. In each row, the cell fragments gradually fuse into linear air channels with cuticular walls and trabeculae. Development of the cuticle walls is started with the formation of particles at the plasma membrane of the aligned cell fragments. Many of the aligned precursor cells deteriorate, leaving space for fluid in the primordial hemolymph channels. Some of these cells survive and become pillar-type trabeculae and a hypodermal layer for replacement of the air sac cuticle in the molt [2, 49].
In the sac-like lamellae of both book gills and book lungs, the hypodermal cells have typical epithelial cell polarity with cuticle produced from the apical surface while their basal surface is in contact with the hemolymph [2, 42, 43]. As is common in invertebrate tissues , no basement membrane is secreted between the hemolymph and basal surface of the hypodermal cells (Figures 3, 7, 8, 10, 11).
The repeated and parallel pattern of the lamellae in book gills and book lungs is evident among the very first lamellae produced in embryos of horseshoe crabs (Figures 5, 9, 12B)  and scorpions [2, 49]. As pointed out above, the basis of the spacing between lamellae is clearly different for the two types of respiratory organs. In spider and scorpion embryos, the air sacs are produced at the apical surface of the precursor cells previously aligned in parallel rows. This results in an air sac between each double row of cells, as evident in earlier light and electron micrographs [2, 14–18].
In the formation of book gill lamellae, there is no initial alignment of precursor cells into parallel rows nor lamellar separation of aligned cells into double rows . Rather, the primordial hypodermal cells apparently proliferate and migrate outward from a small fold or cluster (Figures 5, 6, 9, 12B), producing the opposed epithelial and cuticular walls of the lamellae (Figure 7). This results in some variation in the distance between gill lamellae and a varying number of hypodermal cells in the space between the lamellae (Figures 9, 12B). Thus, many page-like lamellae are formed for both book lungs and book gills, but there are substantial differences in cell activity for lamellar spacing and development in the two types of respiratory organs.