14 A bulk sample (about 75 μm grain size), examined under a petrological microscope prior to XRD analysis consisted mainly of light-brown rhomboidal crystals and fragments lacking secondary oxidative alteration. The peaks of the XRD diffractogram (Fig. 3), however, could not be matched with those of the pure phase of FeCO₃, but with one of atomic substitution of Ca⁺² and Mg⁺² for Fe⁺² had taken place (e.g., Deer et al. 1992). Corresponding to that prominent peak is the empirical formula (Ca_0.1Mg_0.33Fe_0.57)CO₃; minor components in the bulk sample were probably contaminating silicate minerals, or detrital quartz grains from the enclosing shale.

15 The mold 4-Z41b is characterized by four broad lobes separated by narrow longitudinal grooves that converge dorsally without observable remnants of a peduncle or stalk. The larger left-side lobe (Fig. 2) shows curved rows of paired, rectangular prepollen sacs, clearly seen in Fig. 4, that radiate from the dorsal surface to the ventral surface. Halle (1933, p. 46) called such rows “strings of pearls”. The prepollen sacs are smallest in the dorsal region and increase in size ventrally. The marginal areas appear smooth, although obtuse termini can be seen faintly; the presence of these termini is confirmed in a coalified fragment that shows them forming a crenate rim with non-branching uniseriate hair about 60-μm-long (Fig. 5a). Hair typically occurs on the dorsal face of the organ (Schopf 1948, figs. 14C, E–F). The grey-white patches in Fig. 2 show areas on the outside of the mold where siderite is exposed due to the destruction of the compression during coal-mining operations.

16 Study of the compression in cross-section revealed a layered nature. Two layers were observed, occasionally differentiated by the presence of protruding rectangular-rotund terminal prepollen sacs in one layer and their absence in the other. This is demonstrated by the smoother surface shown in Fig. 5b; this is probably the dorsal surface, although the hair is not observable. Fig. 5c shows the obverse, or ventral surface, which is marked by imbricated prepollen-sac termini, faintly visible on the lower end. From the perspective of compression, the fragment, which is about 12 mm wide and 8 mm long (Fig. 4) probably originated from the larger lobe shown in Fig. 2. Illustrated in Fig. 4 are two rows of doubly paired, contiguous prepollen sacs with in situ sporangia and grains that correlate to the pits in the studied mold (see Halle 1933). The fragmentary row consists of five pairs, where the rectangular prepollen sacs are 1500−1900 μm long and 750−880 μm wide. It is significant that these prepollen sacs retained their three-dimensional configuration as a result of the stabilizing influence of siderite during mold formation. Furthermore, because the prepollen grains were not shed at the time when the organs dropped off, they are assumed to be developmentally immature, implying a size bias (Table 1).

17 For description and discussion purposes, the maceration products from three connected prepollen sacs are subdivided into cuticles, acellular surfaces, prepollen sacs, and prepollen grains. As a result of the three-dimensional preservation, in situ sporangia could be observed, which facilitated the separation of opposite surfaces of a prepollen sac (Figs. 6a–b). These surfaces are shown to be acellular, dense, and interlaced with vascular bundles that could not be fully mapped because of the small number of samples available.

18 The compressed non-prepollen-sac material is composed of more than one layer; layers could be separated only with difficulty by the “teasing-apart” method used for separating foliar cuticles (Cleal and Zodrow 1989). Loose fragments of surfaces could be separated on the presence or absence of stomata, but the positions of the surfaces in relation to the upper/lower prepollen-sac surfaces (i.e, in terms of the internal structure) could not be ascertained. In some cases, a surface has small, round, cell-like features about 10 μm in diameter, with a fragmentary stoma preserved by three neighbour cells that are about 12 μm long and 8 μm wide (Fig. 6c). In contrasting cases, a cuticular surface (Fig. 6d) shows well-preserved, somewhat rectangular cells ranging from 17 × 33 μm to 33 × 47 μm (n = 8), with straight and prominently cutinized anticlinal walls. A cyclocytic? stoma from that same cuticle (Fig. 6e) has small guard cells 17 μm long and 7 μm across, surrounded by five distorted wedge-shaped neighbour cells that are 27−50 μm long and about 34 μm wide. Stomata associated with acellular surfaces includ the cyclocytic type (Fig. 6f, enlarged in Fig. 6g). The guard cells are about 10 μm long and 5 μm wide, and the three recognizable and elongate neighbour cells are 40−43 μm long and 20−26 μm wide. The stoma (Fig. 6h) has guard cells about 40 μm long and 17 μm across, and a discernible faint circumscribing oval ring (6−7 μm wide) that could be interpreted either as a stomatal flap or as the bottom tube of the guard cells. This ring is surrounded by faintly visible rectangular to oval neighbour cells that, though visible are 33−40 μm long and 17−27 μm wide. Although only five neighbour cells could be reliably counted, there may be up to twelve. Stomatal density, stomatal index and polar trend could not be determined.

19 Each prepollen sac contains at least six, and possibly twelve, layers with intact grains, three to four across (perhaps in tetrads). Each layer is separated by thin acellular material with delicate vascular bundles, the latter being the most abundant maceration products encountered and probably parenchymatous tissue (Stidd 1981). Among the sporangia recovered, the maximally length is 2 mm; they contain circular and ellipsoidal prepollen grains (Fig. 7a). Figure 7b shows a rare example of a circular prepollen grain, 460 μm in diameter. At the commissural deflection, branches (17 μm long) of a triradiate mark are clearly visible in the water-inflated samples. Figure 7c shows an ellipsoidal grain that is 520 × 380 μm. Table 1 summarizes grain-size data. Noteworthy is the 1:2 ratio of circular to ellipsoidal grains. The distal surfaces of both prepollen forms are marked by deflected commissural apertures with terminally located triradiate rays 10−13 μm long. The proximal surfaces, show vaguely perceptible symmetrical ridges; however, we regard these to be the result of shrinkage and thus not of taxonomic relevance (Zodrow et al. 2017a). Prepollen grains clearly show an alveolate surface structure when immersed in water and magnified at ×250, (e.g., Fig. 7d).

20 The KBr spectrum of the compression part of the specimen (Fig. 8) shows a relatively high contribution of aliphatic stretching moieties in the 3000−2800 cm^-1 region, and distinct bands in the aliphatic bending region, especially the one with the peak at 1446 cm^-1 wavenumber. A prom-inent band of aromatic carbon at 1603 cm^-1 wavenumber and a very small shoulder at 1701cm^-1 represent oxygenated groups. Aromatic hydrogen bends are also present, and they are especially distinct in the out-of-plane 700−900 cm^-1 region, where all three aromatic bands are present. Hydrogen bends in the aromatic stretching region (3000−3100 cm^-1) are of low intensity. All these qualitative characteristics are reminiscent of vitrinite of high volatile bituminous rank (e.g., Walker and Mastalerz 2004, fig. 5). Semi-quantitatively, the compression is characterized by a higher CH₂/CH₃ ratio (2.78) compared with the vitrinite of high volatile bituminous coal, where CH₂/CH₃ ratios were within a range of 1.25 to 1.32, suggesting longer and less branched aliphatic chains for the compression.

21 Micro-FTIR spectra from a prepollen grain are very different from those of the compression (Fig. 9). The most prominent bands of the 1600−1800 cm^-1 region belong to oxygenated groups. Aromatic carbon at ~1600 cm^-1 is much less distinct and is overwhelmed by neighbour oxygenated groups. Aromatic hydrogen bands in either the aromatic stretching region or the out-of-plane region are almost undetectable, whereas aliphatic bands in the stretching region are of high absorbance. This results in very low aromaticity (0.00086, 0.0031), lower ratio of aliphatic stretching bands to oxygenated groups and aromatic carbon (Al/Ox), and a much higher CH₂/CH₃ ratio compared to that of the compression. We note, however, that these ratios vary significantly between the individual sampling positions on the prepollen grain, as shown in Fig. 9, due to either a different original chemistry or to the influence of secondary processes such as oxidation.

22 A theoretically predictable but heretofore unrecognized state of nodular preservation of a medullosalean male organ is here reported from the carbonate facies of the Canadian Sydney Coalfield. The quadripartite configuration, preserved by siderite in three dimensions, suggests that the campanulum originally had a squat bell shape. Whether this confirms the theory of Dennis and Eggert (1978, fig. 45), involving a four-part campanulum based on independent synangial structures, refuted by Stidd (1990), is debatable. However, Zodrow et al. 2017a, figs. 2 and 14) previously reported three compressed medullosalean male organs with tripartite campanuli with a more elongate configuration from the Sydney Coalfield. This suggests a diversity of campanulary configuration.

23 In terms of the internal configuration, we report fragmented rows of paired prepollen sacs preserved through compression, whose pitted imprints on the sideritic mold are arranged in radial fashion from the dorsal to the ventral surfaces. Disk-like compressions/impressions with radial pits are distinctive of Dolerotheca Halle 1933 (e.g., Halle 1933, p. 47, who cited the specimens illustrated as Renault 1896, pl. 9, figs. 5–6; Schopf 1948, p. 719, 721; Cridland et al. 1963, pl. 22, fig. 50; Gillespie and Clendening 1967, pls. 1–3; and Pacyna and Zdebska 2010). In terms of permineralization, however, these rows of paired prepollen sacs (“pit strings”) are equivalent to structures seen in a transverse section of specimens of Dolerotheca (Halle 1933; Schopf 1948, pl. 105, fig. 1).

24 In contrast with the digitiformis arrangement of synangia in the Dolerotheca-like specimens (Zodrow et al. 2017a), the specimen described here is characterized by elongate sporangia. These contain a relatively large proportion of circular prepollen grains relative to ellipsoidal ones, the former commonly showing a triradiate mark related tetrad separation, and thus non-functional, as is the case for prepollen (Schopf 1948, p.701).

25 The scarcity of compression material in the specimen studied, attributed to erosion by coal-mining activity, is reflected in the small number of different cuticular/acellular surfaces observed. In contrast, Schopf (1948) documented a number of differently structured cuticles in permineralized Dolerotheca specimens. The stomatal type identified on the specimen studied compares with that of the Dolerotheca-like organ [see Zodrow et al. 2017a, fig. 6D(2)], and with that reported previously for the coal ball Sullitheca dactylifera by Stidd et al. (1977). Schopf (1948) did not observe stomata. The surface markings, ellipsoidal/circular shapes, and very large size of the in situ prepollen grains fit the concept of Monoletes Schopf 1936 nom. cons.

26 In taxonomic terms, the campanulary shape, structural arrangement of prepollen sacs in paired rows, and the nature of the in situ prepollen grains fit the generic concept of Dolerotheca (Stidd 1981; Rothwell and Eggert 1986; Stidd 1990). However, specific assignment is not warranted based on a single fragmentary specimen.

27 Micro-FTIR results of this study compare closely with the KBr results of the prepollen grains of Dolerotheca-like prepollen organs (Zodrow et al. 2017a, fig. 12). The dominance of oxygenated groups, prominence of aliphatic stretching bands and indistinct aromatic H-C bands both in the stretching and in the out-of-plane region are all characteristic of prepollen. The main difference detected in the qualitative analysis of the spectra is a higher contribution of aromatic carbon in the Dorotheca-like prepollen organs when compared to the prepollen of the current study. Comparison of the functional-group ratios indicates a slightly lower Al/Ox ratio for the current grain (0.20–0.29) compared to the Dolerotheca-like prepollen (0.34), suggesting more advanced oxidation for the grain studied. In turn, a higher CH₂/CH₃ ratio (5.5–8.4 versus ~4 for the Dolerotheca-like prepollen), suggests higher and straighter aliphatic chains for the prepollen studied here. We suggest, however, that these spectrochemical differences could be due to diagenesis (e.g., Lis et al. 2005), although some differences, especially the length of the aliphatic chains, may also have resulted from subtly different taphonomic influences.