Dubious Droplets
Apart from the presence of beads of iron, there were other indications that heat had been involved in the creation of the H-glaze.   Scattered across the surface of the ferrous deposit were what looked like droplets or bubbles of a dark brown semi-transparent material.   Both Levengood and I noticed the broken remains of these indicating that they had once been hollow domes.   They clearly needed to be identified and so, when I sent a sample for examination in a scanning electron microscope, I specifically asked for these features to be analysed. In fact, I went to considerable lengths to point out these anomalies and explain their significance.  I was assured that they were silica.  The implication was that they had formed from molten silica by gasses emitted from the glaze.

However, as I continued my investigation of this material, I began to doubt what I had been told.  After all, the surface of the glaze was covered with small fragments of silica.  Why had they not melted?  Eventually, I returned to the individual who gave me this assurance and asked him to confirm his original diagnosis.  His evasive response convinced me that those features were not silica at all but, most probably, some form of corrosion product from the iron.

However, it was not at first clear how rust could form into a hollow sphere. The following is one possible explanation.  For corrosion product to develop into a spherical shell, it must have some sort of former - something spherical around which the product can develop and grow. The most likely candidate would be a small drop of water.  Corrosion is an electrochemical process and, at some points in contact with the water, the iron oxidises and enters solution as cations - atoms of iron with two electrons (e^-) removed.  These electrons form a small current which flows to the edge of the droplet where it causes     oxygen reduction to occur at the air/water/metal interface producing hydroxide ions.
                                           O₂ + 2H₂O + 4e^- → 4OH^-
These hydroxide ions (OH^-) react with the Fe⁺⁺ cations to form ferrous hydroxide.
                                            Fe⁺⁺ +2OH^- → Fe(OH)₂
However, ferrous hydroxide is not very soluble and precipitates at active points along the edge of the droplet.   As a result, after quite a short period, a shell of hydroxide develops over the surface of the droplet and will eventually completely cover it.   On exposure to air, the colourless ferrous hydroxide readily oxidises to brown ferric hydroxide, Fe(OH)₃.  In time, the water will disperse leaving the hollow spherical shell which slowly converts to ferric oxide Fe₂O₃.xH₂O.
Of course, this process requires special weather conditions.   Too much rain would saturate the iron and no individual droplets of water would form on its surface.  Also, the size of rain droplets is substantially larger than the observed shells.  In this respect, it is noteworthy that Rob Irving has commented on several occasions that a fine drizzle was falling when he distributed his iron powder.  Providing such precipitation was not too prolonged, it would have produced the ideal conditions for the development of the shells of corrosion product as described above.
It is possible to find examples of these bubbles occurring naturally on corroding ironwork.  The example on the left was found on a section of an iron brace cemented into a stone wall.   Note the hole in the old large shell revealing its hollow interior. I suspect that the smaller shells surrounding it were younger.  It is also worth noting that these features occur more often in alkaline conditions because ferrous hydroxide precipitates more readily in such situations. Hence the significance of this example being found near cement and the H-Glaze being found in a field laden with chalk.

As a final confirmation of this explanation, when I was conducting the distribution tests described on the following page, I observed numerous small shells of hydroxide forming on the iron coatings as they rusted.   Each shell was forming around a small droplet of water.

In conclusion, there can be little doubt that the existence of these anomalies is due more to the damp, drizzly British weather than heat generated during some preternatural event.

An important feature of the glaze was its even distribution.   In the following page, I will describe some tests using Rob Irving's iron powder intended to see if this characteristic could be reproduced and under what conditions.