High-voltage (1.0 MV) electron microscopy and stereomicroscopy, electron probe micro-analysis, electron diffraction and three-dimensional computer reconstruction, have been used to examine the spatial relationship between the inorganic crystals of calcium phosphate and the collagen fibrils of pickerel and herring bone. High-voltage stereo electron-micrographs were obtained of cross-sections of the cylinder-shaped intramuscular bones in uncalcified regions, in regions where only one or only several crystals had been deposited in some of the fibrils, and in successive sections containing progressively more mineral crystals until the stage of full mineralization was reached. High-resolution electron probe microanalysis confirmed that the electron-dense particles contained calcium and phosphorus. In the earliest stages of mineralization and progressing throughout the mineralization process, the crystals are located only within the collagen fibrils;​ crystals are not observed free in the extracellular spaces between collagen fibrils. The progressive increase in the mass of mineral deposited in the bone tissue with time occurs, essentially, completely within the collagen fibrils including the stage of full mineralization. At this stage, cross-sectional profiles of collagen fibrils are completely obliterated by mineral. A small number of crystals that are located on or close to the surface of the fibrils appear to extend a very short distance into the spaces between the fibrils. These ultrastructural observations of the very onset of calcification in which nucleation of the calcium phosphate crystals is clearly shown to begin within specific volumes of collagen fibrils, and of the subsequent temporal and spatial sequences of this phenomenon, which shows that calcification continues wholly within the collagen fibrils until maximum calcification is achieved, add important information on the basic physical chemical mechanism of the calcification and the structural elements that are involved.

The spatial and temporal independence of the sites where mineralization is initiated establishes that such ultrastructural locations within individual collagen fibrils represent independent, physical chemical nucleation loci. The findings are totally inconsistent with the proposal that crystals must first be deposited in matrix vesicles, or other components such as mitochondria, and subsequently released and propagated in the interfibrillar space, until they eventually reach and impregnate the hole zone regions of the collagen fibrils.

Three-dimensional computer reconstruction of serial transverse and longitudinal sections demonstrates periodic swellings along the collagen fibrils, corresponding to the hole zone region of their axial period as mineralization proceeds. Bulging of fibrils was observed only in the more heavily mineralized regions of the tissues, a result suggesting that structural changes of the collagen fibrils may occur to accommodate additional mineral mass during the progressive increase in the deposition of a solid phase of calcium phosphate within the fibrils.

We have also examined the same fish bones used for high-voltage electron optical studies using neutron diffraction techniques, from which the volume of space within collagen fibrils potentially available for the deposition of a solid mineral phase of calcium phosphate can putatively be calculated. Contrary to the direct visual observations of the location of the calcium phosphate crystals by high-voltage electron microscopy, calculations based on the data obtained from neutron diffraction techniques reveal that there is not a sufficient volume of space within the collagen fibrils of fish bone to accommodate all of the mineral phase present in the fully mineralized tissue. This discrepancy between direct observation by high-voltage electron microscopy and quantitative calculations of the three-dimensional space within collagen fibrils based on neutron diffraction and theoretical models of the packing of collagen molecules in collagen fibrils, raises serious questions about the validity of the particular neutron diffraction methods employed, and the assumptions used to calculate the intrafibrillar volume of the collagen fibrils. The correctness of the high-voltage electron microscopy studies is confirmed in other experiments in which the extracellular, interfibrillar volume of space in the tissue was measured directly by high-voltage electron microscopy of cross-sections of the tissue. The interfibrillar volume was found to be very much less than that required to house the large proportion of the mineral phase in the tissue calculated to be present in the interfibrillar space of fully mineralized bone from the neutron diffraction data using the methods and assumptions of Bonar and co-workers.