The Trindade Island Photographs 16 Jan 1958

Martin Shough

A study of the ship's position and orientation
(Mar 2004. Last update 20 Mar 2004)


There is general agreement that early press accounts which placed the Almirante Saldanha near the northern tip of the island, off the Pta. Crista do Galo (Galo Point), were very inaccurate. Not only is this a long way from the landing at Praia dos Portugueses (Portuguese Beach) on the east side of the island where the Navy Oceanographic Post was located, but also this is proved by triangulation based on the island topography.

Kentaro Mori has compared the Brazilian Navy Hydrography and Navigation Division chart of the island with Barauna's pictures and with other photographs of the island to conclude that the ship was probably anchored at this position few hundred metres off Pta. do Valado (Valado Point). However he notes that the topographical detail on the chart is known to be unreliable and hence an exact position is difficult to pin down.

Brad Sparks (personal communications) has independently attempted to resolve the same problem by correcting the topographical chart from satellite and other photographs, matching the result with a carefully orthorectified panorama of the island constructed from Barauna's photographs. Sparks' analysis confirms the approximate location off Valado Point, although full details of this ongoing work are still awaited.

Fig.1 Photo from O Cruzeiro, 8 March 1958 (courtesy Brad Sparks)

So these lines of investigation appear to converge. But even with corrections made to the chart the match remains imperfect.

Moreover even the best fit places the ship at a point inconsistent with navigational beacon fixes recorded in the ship's log, unless bearings recorded as "true" are in fact magnetic (Brad Sparks, personal communications).

And why anchor off Valado Point anyway? Why not at the more convenient anchorage charted in Portuguese Bay?

Indeed we have a photograph (Fig.1) which does seem to show the ship anchored in the bay close to Portuguese Beach. If this were a photograph of the ship lying off Valado Point then the line of sight from the camera to the ship would be running significantly west of due north. North would then be broadly-speaking off the right edge of the photograph and shadows should run away from the camera. But the shadows appear not to run away from the camera. In fact they appear to lie in a direction running slightly up the slope of the beach to the left, indicating that the sun is in front of the photographer's right shoulder and thus significantly north of the zenith, which is impossible at the latitude of Trindade in January. On the other hand the picture is consistent with a line of sight roughly NE towards an anchorage off Portuguese Beach, which would place the sun high in the southeast.

Do these problems indicate some more serious inconsistency? We can be confident that the UFO photos really were taken from the deck of the Almirante Saldanha (see A study of the camera position) during the January 1958 visit to Trindade. But if the ship was in fact not anchored in the position it appears to be from Barauna's photos, then (assuming that the anchorage position was not moved, and reportedly the log indicates no such movement) the ship must have been underway at the time the photos were taken. And if they were taken before dropping anchor, or after weighing anchor, this would probably be in serious conflict with the testimony.

So it will be useful if our own investigation can act as a check on the results of Sparks and Mori, as it were from a third perspective. We begin from the unpromising photograph shown as Fig.2, which was published in the same issue of O Cruzeiro as was Fig.1. This picture shows Almiro Barauna himself on the island with the ship in the background, and in this case, we will show, the photogrammetry is consistent with a more northerly position at anchor off Valado Point.

Fig.2. Barauna at the Trindade Island Oceanographic Post on Praia dos Portugueses in Jan 1958. Almirante Saldanha at anchor in the background. (Printed in O Cruzeiro, 8 March 1958. Thanks to Kentaro Mori and Brad Sparks)

The Brazilian Navy Hydrography and Navigation Division chart of the island is Chart 21, issued 1938 with major revisions to 1971 and minor revisions to 1999 (shown here, courtesy Kentaro Mori). As mentioned it is obviously approximate in many places and is known to contain inaccuracies in the values and locations of some spot heights, not so surprising given the extraordinary complexity of the landscape, which has been described as "a topological disaster". At least the coastal features, and especially the region around the charted anchorage in the Enseada dos Portugueses (Portuguese Bay), ought to be dependably surveyed; but the exact location of the Oceanographic Post sign shown in Fig.2 is impossible to infer from this chart. Even the layout of buildings may have been quite different in 1958.

Nevertheless the general location of the installation is on or behind Portuguese Beach, which is consistent with Fig.2. The few buildings marked are set back off the beach a little uphill. Unfortunately these are scattered and the contour heights are not clearly identified, but the closest structure is somewhat less than 50m or about 150' from the shoreline (presumably shown MSL; tide variation is probably only a metre or so) on a rising slope, probably only a few metres above MSL. We can show that the location of Fig.2 is also slightly above sea level because it is possible to discern that the ship's waterline is below the horizon line. The height at which the horizon line (effectively at infinity) cuts the ship's hull gives the approximate height ASL of the camera. The elevation drawing of the Almirante Saldanha shows a height of about 22-23 feet (6.8m) from waterline to deck level, and the horizon line appears to cut the hull roughly in half, so we can say that the camera height is in the order of 10' ASL.

Above: Galo Point from Barauna's photo P1

Below: Detail from Fig.2, taken from slope behind Portuguese Beach, Jan 1958

Below: Photo taken at sea, SE of Portuguese Beach and from a shallower angle, with high contrast detail of Galo Point shown at right. Compare image top right.

Fig.3. The landform on the far left of Fig.2 can be identified as the claw-like pinnacle of Pta.Crista da Galo.

Of most importance in Fig.2 is the feature that enables us to estimate the approximate angular scale of the photograph - the distant pinnacle and stepped ridge on the far left, which we can identify (Fig.3) as the Pta.Crista da Galo, or Galo Point at the northern tip of the island. The ratio (~1:1.27) of the height of the pinnacle (ASL) to the height of the lip of a distinctive square step in the ridge above (less square in Fig.2 owing to the different perspective) confirms this identification.

(Note that the axial ratio of Fig.2 is the same in scanned copies from two different sources, arguing that no horizontal or vertical distortion has been introduced. This is confirmed by examining an image of the complete page in which the halftone is embedded. The text appears undistorted and the ratio of horizontal and vertical dimensions of the text area, 1.4:1, is consistent with the I.S.O. standard metric paper system, later adopted under B.S.I. standard B.S.3176 of 1959 as the 'A' series. The simple optical and mechanical processes of halftone printing in 1958 would not afford any opportunity for axial distortion. The original photograph might not have a perfectly flat field, especially at the periphery. Arguably the detail of Galo Point from Fig.2 appears as though it might be laterally compressed.


One check is the ratio of mast height to mast separation. Given a flat field with the ship broadside-on the height of the four masts should be 81% of their fore-aft separation, which is at least not inconsistent with the fuzzy image enlarged in Fig.4. So no such compression is obvious within 6-7° of the right edge.

But the printed FOV of only about 27° [see below] suggests that this picture might be a cropped detail, in which case the ship could conceivably be close to the optical axis with Galo Point at the periphery of a ~ 40° negative FOV subject to a significant aberration. On the other hand so much aberration seems unlikely for a camera of any quality, and the perspective change seems the most likely cause of an apparent lateral compression. In any case the dimensions of Galo Point that are of interest to us are height ASL, and ratios of heights ASL, which will be negligibly if at all affected by any small theoretical distortion acting radially to the optical axis and thus at a right angle to these dimensions.)

Galo Point and the Obelisk from sea level at Portuguese Beach (courtesy Kentaro Mori) shown as b/w silhouette at right.

Fig.5. The 'claw-like' profile of Galo Point rises over the Valado headland as seen from the shoreline. Compare Fig.2, taken from slightly higher elevation above the beach.

The distance from the Navy installation on Portuguese Beach to Galo Point is approximately 2770m (~1.72 miles), choosing for distinctness a point on the slope just below the structure nearest the shore where sightlines to Galo Point and to the ship would cross a small rocky inlet in the foreground, similar to the feature visible behind Barauna in Fig.2. This point would be perhaps a few tens of metres inland and a few metres higher than the location of the camera in Fig.5. From here the elevation of the summit of Galo Point above the sea horizon allows us to calculate a fairly reliable angular yardstick from which we can determine the proper location of the ship - provided we know the height of Galo Point.

Unfortunately, in the absence of an on-the-spot survey this is not known independently because it is not shown on the chart. But all is not lost because the constraints on position coming from triangulation of the Barauna panorama yield a first-order approximation, and from the relatively well-known angular scale of the Barauna prints a value for the height of Galo Point can be deduced. Our calculation will therefore become an independent angular test of the internal consistency of the triangulation leading to that approximation.

Fig.6. Drawing based on Chart 21 of the approximate angular relationships in Fig.1.

The distance from the Galo Point pinnacle to the position inferred by Mori is a distance of some 1650m. If we take the FOV of Barauna's "6x6" format Rolleiflex 2.8E to be 38° wide and the print area to be negligibly cropped (this is consistent with the figure derived from rough photogrammetry to establish the shipboard camera position and with the optical formula for a ~55mm negative width and 80mm FL). The measured angular height of Galo Point is then 5.9°, corresponding to a linear height of 170m at 1650m.

Transferring this result to Fig.2 we have 170/2770 = 0.61 = tan 3.5°, giving (with Galo Point measuring 22mm high on picture dimensions 168 x 98mm) an angular scale of 6.2 mm/1°. The fore-and-aft angular subtense of the four masts of the Almirante Saldanha in Fig.2 can now be directly measured at 3.17°, and the actual linear distance between them can be measured from the 1/120-scale ship's plans to be 170' (51.83m) giving a limiting range from the camera of approx. 51.83/tan3.17° = 942m for the case that the ship is orientated normal to the line of sight.


A crude check can be made on Fig.7 (from which the detail in Fig.5 is taken). From Chart 21 the angle subtended at the Navy post on Portuguese Beach by Galo Point and the unnamed peak with a spot height of 309m can be measured as approximately 21°. The sea level at Galo Point is not shown but it can be calculated from photos like those in Fig.3, where the distance from the tip of the pinnacle to the sea is found to be 3.17 times the distance from the tip of the pinnacle to the very distinctive square-profile ledge above it. The angular height of Galo Point can then be measured off as approximately 3°. Galo Point can then be 160m high at a distance from the camera of 3077m, both figures within 10% of our other estimates which, given the gross approximations involved and the uncertain precision of Chart 21, is reassuring.

So our result places the ship about 940 m from the estimated Fig.2 camera position, within some 180m of Mori's triangulated position as illustrated in Fig.6. This is a crude procedure that can't in any way be appealed to in order to refine the triangulated position; it is merely a test of the consistency of that position with independent data. The triangulation passes the test. There is no doubt that the Almirante Saldanha was anchored off Valado Point during this Jan 1958 visit, within some 300m of the Valado Point beacon, in a position consistent with the UFO photographs.

But this result is definitely not consistent with the photograph of the ship at anchor off Portuguese Beach in Fig.1. The photo in Fig.2 is certified as to date because it shows Barauna on the island: "Almiro Baraùna, autor das fotos, quando disembarcou na Ilha da Trindade, vendo-se ao fundo o navio da Armada 'Almirante Saldanha'". On the other hand Fig.1 is captioned as "O Almirante Saldanha, agora navio hydrográfico, a serviço do ano geofisico internacional, ancorada ao largo da Trindade". This means that the date may be uncertain because the ship had visited Trindade before for International Geophysical Year activities. The IGY, conceived in 1952 (by Lloyd Berkner) and five years in the planning, was actually 17 months long, running from July 1957 to December 1958 and involving 67 participating nations. As part of Brazil's commitment the Almirante Saldanha was first lent to the Navy Hydrography & Navigation Department (DHN) in April 1956. She was was partially refitted for oceanographic IGY work in January 1957, sailing from Rio in February, and in August 1957 she was officially assigned to the DHN. The Trindade installation - the base of anti-submarine operation abandoned after the 2nd World War - was re-opened as a meteorological station and began observations from October 1957. It is therefore possible that Fig.1 was obtained by O Cruzeiro from another Navy source and originates from a 1957 visit when the Portuguese Bay anchorage was used (limits on the solar elevation determined from shadows could maybe narrow down the possible time of year).

One other issue is worth mentioning. The Portuguese Bay anchorage can be seen to be marked on Chart 21 by the intersection of lines of position (LOPs) from pairs of range beacons on Portuguese Beach and on Pedra Point. The navigational utility of such beacons is obvious. Fig 6 shows that the 137° line of position to the beacons on Pedra Point also happens to intersect rather accurately the estimated January 1958 position(s) of the Almirante Saldanha off Valado Point, whilst the southernmost position happens to lie more or less exactly due east from the Valado Point single beacon. The situation is shown at larger scale in Fig.8 below.

Fig.8. The red spot 927m from the estimated camera position used to construct Fig.6 marks the intersection of a bearing due E from Valado Point beacon and the charted 137° line of position (LOP) to the pair of range beacons on Pedra Point. (The latter LOP also intersects a LOP from the Portuguese Beach range beacons to mark a charted anchorage point on the modern map.)

However this is almost certainly a red herring. Brad Sparks (personal communication) reports that there is some doubt as to whether the Pedra Point beacons marked on Chart 21 were there in 1958, and he further points out that the ship's log does not cite bearings to Pedra Point. In fact the log consistently cites bearings to Portuguese Beach and Valado Point over the course of three days. This suggests that the positioning of the ship on the Pedra Point LOP can only be a coincidence.

I do not have access to the ship's log, but Sparks reports that the position fixed by the bearings given in the log is consistent with the position fixed by triangulation from Barauna's photographs and with the position independently inferred here (within the margins dictated by topographical uncertainties and the variability of position due to scoping of the ship's anchor chain), provided that the bearings logged are taken as 'magnetic'.

© Martin Shough March 2004

(back to Trindade index)