At great distance, on mainland Brittany, any such object(s) would have to be in the region of 2nmi (4km) across to subtend the angle observed. Even allowing for a very large error in visual angle estimates, no candidates (apart from the lakes) appear to exist. This leaves us with objects on the island of Guernsey. Realistically the only large reflective objects on Guernsey are the flower and vinery glasshouses. This of course was Capt Bowyer's very first interpretation of UAP#1 because the initial LOS passed approximately across the middle of Guernsey and he had seen specular reflections from the glasshouses before, lasting a few seconds.
Such reflections would not be surprising. Sunshine on Guernsey was abundant (p.69). There are many commercial glasshouses on the island extending in all to 1.539 x 106m2 of glass (2.54% of the island area), in addition to many polythene tunnels (map, p.78). They are built in blocks often 4-5,000m2 in extent or more, such as those of the Guernsey Clematis Nursery in Vale (10,000m2 and 4,450m2 ) and St Sampson's (5,750m2). The Vale buildings are typical of most of the glasshouses in Guernsey and are of the "Venlo" design (Fig. 29). Ridge orientations are very variable. The ridges at Vale all run approximately E-W, the largest block being 200m x 50m n1 Emails from Paul Ingrouille, Production Manager, The Guernsey Clematis Nursery Ltd, St. Sampsons, Guernsey, to J-F Baure, 06..07.07 & 14.07.07; Dave Killan, Director, Digimap (Jersey) Ltd to Paul Fuller, 29.02.08, 06.03.08. .
In this case the Venlo ridge angle of approximately 23° is close to the optimum angle for specular reflection given the elevation of the sun at the time of the observation, and it seems quite plausible that one of the glasshouse blocks might have been oriented so as to direct specular sun reflection towards the Trislander. The expected ratio of angular sizes (1.86) would be closer to the visually estimated ratio (2.6) than was the case for much more distant Breton bays or lakes.
Even a 200m length of glasshouse subtends little more than ~6arcmin from the range of the initial sighting ~52nmi away, and even at closest approach (~28nmi) would subtend only ~12arcmin. Thus there is a factor 5 discrepancy in angular size from the witness estimates of 0.5 - 1.25°. But this need not be a problem if we assume reflections from multiple adjacent blocks, with a non-reflective strip of ground between them perhaps accounting for the "dark band"?
Of more importance is the persistence of the "sharply-defined" images of UAP#1 and UAP#2 for some 12 minutes and 6 minutes respectively, despite changing reflection geometry, remaining identical in appearance during level cruise at 4200ft and during about 2000ft of descent. The motion of the aircraft represents about 0.5° change in elevation (1.2° to 0.75°) relative to a reflector on Guernsey, which is not large. Given a planar reflecting surface we would expect the angle over which specular reflection occurs to be comparable to the angular diameter of the sun (0.5°). The change in bearing required is an order of magnitude larger than the solar diameter, which is more difficult to accommodate, seeming to require a reflector with a radius of curvature rather than a plane reflector like a glass roof.
Fig.29. Typical "Venlo" style glasshouse on Guernsey with 2m x 1m glass panes laid to a ridge angle of ~23°. Other "Q22" glasshouses have 75cm panes inclined at ~30° and are built in multiples of a 7m span. (Photo courtesy Paul Ingrouille, Guernsey Clematis Nursery Ltd.)
We note that an observer familiar with specular sun reflections from Guernsey glasshouses, in varying conditions during hundreds of near-identical trips on this airway over the course of 8½ years, rapidly discounted the theory. The "escalation of hypotheses" happened initially because UAP#1 failed to disappear as a reflection normally would. For Capt Bowyer, and for us, the strangeness of that fact was reinforced by these further factors:
First the pointed cigar shape, edge-definition and "dark band" observed through binoculars; secondly, the appearance of the smaller but otherwise identical UAP#2 offset to the west; thirdly, the independent horizontal angular motion of these two UAPs by several degrees relative to one another; and finally the impression that the angular elevation of the UAPs rose towards zero degrees just before disappearing as the aircraft descended to the 2000ft haze layer.
A distorted mock mirage of Guernsey glasshouse reflections could not explain images observed at changing elevations, varying from a significant depression angle below the horizon ("2 degrees", "against the sea and the land", "on the sea") to near zero degrees at the top of the haze layer as the Trislander descended to the haze. We believe that the exit angle from the duct would be tightly constrained to within a few minutes of arc and could not vary in this way n2 Email from Andy Young, San Diego U. to Martin Shough 31.08.07. "Ducted rays cannot be inclined to the local horizontal by more than 1.4°, even if the hot air above the inversion has infinite temperature. For likely temperature differences such as the 5° we have been discussing here, the maximum inclination of ducted rays is only a few minutes of arc.". And more importantly no vertical temperature structure can explain a lateral image displacement of several degrees (or indeed 5x lateral magnification) - still less can it explain steady, independent lateral motions of these displaced images.
In any case, according to the Meteo-France computer simulation and consistent expert advice from Channel Islands meteorologists (Section 5) a localised duct over the Breton coast has broken down some distance south of the Channel Islands area, leaving only a weak inversion (2-3°C/kft, ~1/10 ducting strength) in the Channel Islands area.
To these problems we can add that the triangulated apparent location of UAP#1 at >1500ft near Alderney was intersected by the reciprocal sightline from Capt Patterson to an unusual yellow object at a similar height in a similar location, which was over the sea, far from Guernsey.
Plausibility (0-5): 2