*Meta Research, P.O. Box 15186, Chevy Chase, MD 20825-5186
†Pacific Sierra Research Corporation, Arlington, VA 22209-2369
‡ Univ. of Tennessee Space Institute, Tullahoma, TN 37388
& Parsons Infrastructure & Technology Inc., 4055 Executive Park Drive, Cincinnati OH 45241
% Hernandez Engineering Inc., Houston, TX 77062
* Northern Maine Technical College, Presque Isle, ME 04769
In science, it is axiomatic that "extraordinary hypotheses require extraordinary evidence". This is not a precept, but an observation about the progress of scientific knowledge. Evidence that refines or extends knowledge in a way that is consistent with accepted beliefs is rarely scrutinized to dispel skepticism about its validity, since there is ordinarily no cause for such skepticism. In contrast, evidence that suggests contradiction to existing beliefs will be viewed with a degree of skepticism proportional to the cost of accepting such evidence. That cost, both psychological and economic, of changing direction, i.e., of discarding widely held beliefs, is too great to be accepted without the overwhelming pressure of evidence that cannot be credibly disputed. Thus, evidence that would necessitate such a change of beliefs is, and must be, subjected to a level of scrutiny that would be inappropriate (cost-ineffective; unjustifiably tedious) if applied to evidence consistent with accepted theory. Evidence that survives this extraordinary scrutiny is extraordinary evidence.
Here we will apply the a priori principle. A posteriori findings are of unknown or ambiguous significance almost regardless of their probability of occurrence. By contrast, the probability of occurrence of a priori predictions is as significant as that probability allows. When hypothesis-distinguishing a priori predictions are fulfilled at a level where chance is an unlikely explanation, we normally conclude that the hypotheses are in fact distinguished by the data. In most applications, this principle is invoked to deny statistical significance to a posteriori findings that might appear highly unlikely to arise by chance.1 This is because (like winning a lottery) the probabilities are only low for predicting a specific outcome (the particular winning ticket holder) before the results are known, but are generally not low for some unusual or surprising outcome (e.g., the view of the winner after the results are announced). Therefore, when many improbable outcomes are possible, no special significance may ordinarily be attached to some of them happening. Here we show a finding that robustly survives this principle's application because the occurrences were both the fulfillment of a priori predictions and of extraordinarily low probability to arise by chance.
In 1976, the Viking 1 and 2 spacecraft photographed much of Mars at a medium resolution of about 50 m. per pixel. In 1977, Vincent DiPietro and Gregory Molenaar found a second of two similar pictures of the so-called "Face" on Mars in the region named Cydonia at west longitude 10°, north latitude 41° (the first one shown in Fig. 1 was found by Tobias Owen). The resemblance of the object’s sunlit portion to a humanoid face has aroused widespread attention from both scientists and tabloids. However, the number of random shapes formed by structures on moon and planet surfaces is so vast that no statistical significance can be attached to any such finding because of it’s a posteriori nature. Chance “faces” can and do arise in clouds and on landscapes.
Nonetheless, based mainly upon the non-natural appearance of this object and several of its immediate neighbors, two opposing hypotheses were immediately formulated: (1) the object is an artificial structure built by an extraterrestrial intelligence; (2) the object is a natural structure having a face-like appearance produced by tricks of light and shadow. Among the eight tests2,3 originally conceived to distinguish the two hypotheses were these:
Van Flandern3 recently found that the “Face” [40.89°N, 9.52°W] was both located on the equator and oriented upright with respect to the most prominent long-term former location of the Martian geographic pole as determined by Schultz [45°N, 160°W]8, to a computational precision of better than one degree in each angle. This precision is assumed somewhat fortuitous, given that the old pole location was only specified to the nearest 5° in each coordinate (i.e., with a mean error of ±1.7°). Nonetheless, this finding apparently reversed the results of the cultural significance tests in favor of artificiality at better than the 99% confidence level. Combined with the bilateral symmetry test, the odds against a chance natural origin were then 107 to 1. Van Flandern3 makes an argument that the last sudden pole shift on Mars may have occurred 3.2 million years ago, suggesting that the "Face" mesa is at least that old.
The tendency of the results of all known objective tests to favor an artificial origin has resulted in the publication of several books on the Cydonia region in general, and the “Face” object in particular. Among those, four written by professionals are especially relevant.2,7,9,10 Additionally, two peer-reviewed papers on anomalous placements of mound-like features in regions near the “Face” are relevant.11,12
New MGS images of "Face" mesa
Cognizant of this and other information, on March 26, 1998 NASA announced in a press release that the Cydonia region would be one of the prime targets during the month of 1998 April for the MGS spacecraft. Its high-resolution camera is capable of nearly 1-meter resolution if pointed straight down from normal orbital altitude. On April 5, a high-resolution strip photograph roughly 4 km. wide and 44 km. long captured the 1.5-km. wide “Face” mesa from a slant range of about 430 km. The resolution achieved was roughly 5 m. per pixel -- half the resolution capability of the camera at that range. The grayscale range was heavily restricted, with most of the visible side of the Face mesa represented by just four grayscale levels out of a possible 256.
The following day, the raw strip image was retrieved and posted to the Internet,13 and a processed extract of the “Face” mesa alone was released to the media by the Jet Propulsion Laboratory (JPL)’s Mission Image Processing Lab (Fig. 3) It has come to be known as the "footprint" image because the high shadow-producing central facial features seen in Fig. 1 appear smashed as by a giant foot. The reason is that it utilizes even fewer grayscale levels than the raw data, making features virtually invisible from lack of contrast. This is apparently the consequence of the use of excessive high-pass filtering of the image.13 Moreover, JPL provided no information about the unfavorable viewing and lighting conditions under which the image was acquired. The 45-degree emission angle (tilt of the camera line of sight from the vertical) allowed only a bit more than half of the western side of the Face mesa to be seen clearly. The eastern side was hidden by shadow in the highest-resolution Viking images of the Face. The sun was to the southeast of the Face mesa, so if this landform is actually a representation of a face, it was illuminated from the under its "chin." Such a lighting geometry can radically distort the appearance of an actual human face.
Even with proper image processing, an artist’s sculpture crafted to look like the face from the Viking image was not recognizable as such under MGS viewing and lighting conditions (Fig. 4).14
Figure 5 shows the MGS image of the “Face” mesa with normal image processing, but stretched in the direction of the camera line of sight to compensate for foreshortening caused by the camera’s 45-degree emission angle. This stretching operation gives a better view of the foreshortened features on the eastern side of the landform, but does not show a true picture of what the landform would look like if viewed from directly overhead. The eastern and western portions of the landform were separately processed so details in both would be comparably visible. No filters were used. The inset shows the same view with brightness and contrast optimized for the near (west) side only, and marks the apparent facial features seen, starting with those previously identified in Viking imagery. That the height of the “nose ridge” is greater than anything on the west side of the “Face” mesa can be ascertained by parallax between Viking and MGS images. This implies that some of the east half of the “Face” will be hidden behind the “nose ridge”.
Fig. 6 shows the MGS image of the “Face” mesa with advanced image processing to correct the lighting and viewing angle to match Viking conditions (Sun in WNW, spacecraft overhead). The correspondence between the Viking and MGS images is now obvious. An animation showing this transition in steps is available on the web.15
The most visually striking characteristic of the mesa is the almost perfect symmetry and regularity of the mesa enclosure. It is perhaps 200 m. high (about half the maximum height of the mesa), with linear top and linear sides forming right angles, and a smoothly rounded bottom. Carlotto’s stereo pair (Fig. 7) was created from the MGS image with the assistance of photoclinometry from the Viking image. This allows a view of how genuinely anomalous this mesa is in three dimensions, and how out of context it is against its flat, featureless desert surroundings. Image rectification by Carlotto16 confirms that the "nose ridge" is in fact centered on the axis of potential symmetry of the mesa to within 2%. Although we do not include any of these symmetries in our statistics because of their partially a posteriori nature, they are examples of how the mesa displays less randomness at high resolution than it did at low resolution.
If and only if the mesa was constructed to portray a humanoid face, secondary facial characteristics not seen in the Viking photos would be expected in the MGS image. The most conspicuous of these is the shallow triangular prism-shape seen clearly in the original MGS image despite low contrast,13 located directly over the west eye socket (Fig. 8). It is striking to find a triangle with three such joined linear sides ending in vertices in any natural object. But this object has a shape that yields the impression of a furrowed eyebrow. The best-known example of a triangular-shaped feature elsewhere in the solar system is the roughly equilateral-triangle-shaped interior of the Ukert crater near the center of the Moon's visible disk. But in that case, higher resolution Clementine images showed irregular sides and rounded apexes, as expected for a natural feature. Using the same basis as earlier for face-like shapes, we conservatively estimate the probability of an object with such a good triangle shape appearing anywhere on this mesa at 10-4. Other probabilities are applicable to its specific location within the mesa, relative size and orientation (see table). Not only is each of these probabilities a priori, but there is no background of similar objects from which we can choose just ones that fit our concept of an eyebrow on a face.
Additional secondary facial features make their first appearance in the MGS imagery. In the “eye socket”, at the lower center, we see a circular feature resembling an iris (Fig. 9). The end of the “nose ridge” contains two roundish impressions suggestive of nostrils (Fig. 10) about as closely as one could expect, given so few grayscale levels in this part of the image. Moreover, the ridge appears to narrow as it approaches the forehead, just as a nose does on a real humanoid face. There is even a possible representation of the facial septum joining nose to upper lip, where the “mouth” feature appears to curl upward slightly as on a real face. And the mouth feature appears to have parted lips, another feature widely unrecognized in Viking imagery. The depth of the eye socket and the “ravine” between the parted “lips” ensure that shadows at virtually any Sun angle will create the visual impression of an eye and a mouth, just as they did in the Viking photos. The combined odds against all the "coincidences" in the table arising by chance are 1019 to 1. In combination with our earlier odds, the overall probability of a natural origin is 10-26. This easily covers even the extra-conservative criteria for judging the realistic significance of anomalous phenomena recently advocated. 17
The keys to the validity of this statistical argument are that 1) the secondary facial features exist; 2) there is no background of comparable objects to choose from; and 3) they are predicted a priori by the artificiality hypothesis. Image processing experts (e.g., Carlotto, DiPietro) agree the features seen, although of low contrast (because the original image had so few grayscale levels), are not artifacts of image processing. And any one of the secondary facial features by itself provides a test sufficient to disfavor a natural origin for the mesa. The lack of background confusion of a similar character to these features is evident by inspection. To be an a priori hypothesis, no one needs to state it in print; general agreement that the competing hypotheses do have the stated implications is sufficient. Then the statistical argument can be seen as analogous to being dealt a bridge hand of 13 spades. After the fact, that hand is just as probable as being dealt any other specific hand. But naming the exact hand in advance of being dealt it is so improbable as to merit the conclusion that something extraordinary has occurred. And the same can be said for the appearance of secondary facial features having the right size, shape, location, and orientation on a mesa suspected on other grounds of being a “Face”-portrayal artifact.
Some of these features are inconspicuous when not illuminated by direct sunlight, as in the MGS image. No doubt, erosion and wind-blown sand have taken their toll. And they may have been intended for daytime-only viewing. However, if the mesa is artificial, then artificial lighting from the mesa enclosure might have produced similar effects at nighttime, increasing the contrast for easy viewing from above. But it suffices for hypothesis testing that these features exist, whether or not conspicuous. Many other a posteriori features of the mesa also do not appear random, but decorative, as for a face in costume. Especially noteworthy are the linear features in the “helmet” just above the enclosure top, the symmetric “crest” in the sand just above the “helmet”, the linear markings on the "cheek" and "chin", and the apparent L-shaped right angles in the upper left and right corners of the mesa enclosure.3 Allowing for some erosion, especially on the nose ridge that sticks above the protection of the enclosure, one can picture that everything about this mesa might be part of a planned design. Of course, none of these a posteriori points influenced our probability calculation.
Additional anomalies
The Face mesa is not the only feature difficult to explain as a natural formation. In the April 23 MGS image of the “City” area of Cydonia just west of the Face mesa, we can see a set of imbedded triangles (Fig. 11). Two or perhaps three additional, roughly equi-spaced bases can be faintly traced to the upper right of the most conspicuous triangle. Numerous other features of high regularity, symmetry, and/or angularity also appear in all three MGS images of the Cydonia area, including many more examples of triangle shapes. The Face mesa is far from being an isolated singularity, since many nearby objects also have attributes suggesting artificiality.
Among the most difficult of all of these to explain by natural processes are the "anomalous surface markings" (Fig. 12, noise-filtered in Fig. 13). The largest of these are about 50-60 m. tall, 30-40 m. wide, and are blacker than any other feature in their immediate vicinity. They are located behind the so-called “D&M Pyramid” near the center bottom in the April 5 image, just above a bright "V"-shaped feature that dominates the area. These are fully integrated into the brightness and background noise levels for the image, precluding some sort of accidental superposition in a terrestrial laboratory. Besides, MGS Imaging Team Principal Investigator Michael Malin represents that the image was decoded from bits and converted to Internet-viewable formats entirely inside computers, the process witnessed by reporters, in the span of less than an hour before the image (still downloadable today11) was posted to the Internet and accessed by some of the present authors, among thousands of others. Resembling simple symbols made of lines and curves common to many terrestrial languages, including a horizontal symbol-string arrangement with an occasional taller symbol on the left, these would certainly be regarded as having a human origin if found on Earth.
Discussion
Scientific Method teaches us to distinguish a priori and a posteriori statistical arguments to avoid self-deception, both from falsely attaching significance to chance occurrences, and from failing to attach significance to genuine anomalies. Conservatively estimated, the combined probability of simultaneous chance occurrence of all the features of the Face mesa alone described in this article and predicted a priori by the artificiality hypothesis is 10-26. On that basis, we logically ought to rule out a natural origin for what we see of this mesa. Of course, we have not attempted to estimate the very small probabilities for competing hypotheses such as someone faking or altering the images before they were placed on the Internet. The most reasonable interpretation (in the sense of Occam’s Razor) is that the mesa was built by intelligent beings because that hypothesis requires no changes to known physical laws and no vast, apparently motiveless conspiracies. When additional artificiality indicators seen throughout the Cydonia region are also considered, we suggest that “extraordinary evidence” of the Cydonia artificiality hypothesis now exists at a level that makes it highly competitive with any alternative hypothesis.
Although Levin18 has now concluded that the Viking Labeled Release experiment did in fact detect living microorganisms in the soil of Mars, most astronomers and astrobiologists recognize the improbability of an intelligent species native to Mars. However, that does not affect the likelihood of interplanetary visitors, whose origins and motives we cannot know. The short technological history of our own species strongly suggests that when a species demonstrates the ability to design and build megalithic monuments, the probability that it will not also develop space travel is likely to be small. With the limited knowledge in hand it would be useless to speculate, and imprudent to make speculative assumptions, about the origins of such visitors. The same may be said about the implications of the portrayal of a humanoid face, which may resemble either the builders or the intended viewers, or both.
We can only say with assurance that here is a scientific puzzle of great import. We urge that the highest priority be attached to obtaining additional images of the entire Cydonia region at the highest available resolution. This seems the only way to resolve our mystery with any confidence. In particular, re-imaging the “Face”, embedded triangles, and anomalous surface markings from overhead, under better lighting conditions, with better resolution and contrast, should confirm the existence of the features discussed here and assure everyone they are not the result of tampering or amazing statistical flukes. In our estimation, enough evidence now exists to justify interrupting or extending the MGS mission until the entire Cydonia area can be mapped at high resolution.
SECONDARY FACIAL-FEATURE PROBABILITIES | |||||
Feature | locn | size | shape | orient | bkgnd |
Eyebrow over Eye | 1/35 | (27-10)/54 | 1/10,000 | 30/360 | No |
Iris in Eye Socket | 1/23 | (12-2)/54 | 1 | 1 | No |
Nostrils on Nose | 1/80 | (9-2)/54 | 1 | 30/180 | No |
Lips on Mouth | 1/62 | (28-12)/54 | 1/1,000 | 30/180 | No |
Figure 1. “Face” and “City” at Cydonia on Mars. Excerpt from Viking frame 35A72 of 1976 July 25.
Figure 2. “Face” with dark side detail. Dual contrast-stretched view from Viking frame 70A13 shows detail simultaneously on both sunlit and shadowed sides, suggesting significant bilateral symmetry.
Figure 3. MGS hi-res image of “Face” as released to media. Taken from 1998 April 5 strip image over Cydonia region. High-pass filter virtually eliminated mesa surface features so that even details already seen in all Viking images could not be recognized.
Figure 4. Sculpted face by Kynthia. When viewed with-MGS perspective and illumination, it is no longer recognizable as a face.
Figure 5. MGS hi-res image of “Face”, properly processed. View is from a moderately low angle to the west, illuminated from a low Sun toward the SSE. Dual contrast-stretched to make details for both sunlit/foreshortened (east) and ambient-light/perspective-stretched (west) sides of mesa comparable in brightness. Inset has apparent facial features marked for easier identification.
Figure 6. MGS hi-res image of “Face”, computer-processed to correct lighting from SE to NW and viewing angle from 45° W to overhead – conditions similar to those encountered by the Viking spacecraft in 1976. This is the product of three image-processing professionals, M. Kelly <www.electrobus.com>, M.J. Carlotto <www.psrw.com/~markc/Articles> and B. Starosta <www.starosta.com>.
Figure 7. Stereo view of mesa from ground level. Pairs can be fused without a viewer by relaxing the eyes until they are focused at infinity. This creates four images from the two on the page, one pair for each eye. When the middle pair of images are overlapped, the image should pop out in three dimensions. The symmetry and regularity of the enclosure, and the contrast with its surroundings, become particularly evident in this Carlotto recreation based on Viking and MGS imagery combined.
Figure 8. Close-up of “eyebrow” feature. Despite low contrast, two linear sides can be traced in the “eyebrow”, and a third connecting them can then be traced just above the white, brightly sunlit surface. (Refer to inset.) This latter shiny surface faces toward the Sun low in the SSE, and therefore is roughly perpendicular to the triangular top, apparently on the front side of the “eyebrow” block.
Figure 9. Close-up of “iris” feature within “eye-socket”. Circular feature at bottom center of “eye-socket” resembles an iris. (Refer to inset.) Both “eyebrow” and “iris” features have roughly the right size and shape, and are in the correct position with the correct orientation, to be a deliberate attempt to portray facial attributes. Except for the iris (a circle). no background of comparable objects exists from which the mind can select ones fitting a face model by chance.
Figure 10. Close-up of “nose ridge” feature. Shows narrowing of “nose” toward the forehead, as on a real face. Roundish features at the end of the eroded “nose” appear to portray nostrils, despite minimal available grayscale levels. No similar features appear elsewhere on the mesa. Yet these have the correct size, shape, location, and orientation to be deliberate.
Figure 11. Embedded triangles feature. While of low contrast, three triangles can be clearly traced when viewed with suitable resolution and contrast. At least two more congruent triangles with a common apex at the lower left appear to exist as well. Natural triangles comprised of linear sides terminating in angular apices to this degree are unknown on other moon and planet surfaces. Yet dozens are found in the Cydonia region.
Figure 12. "Anomalous surface markings" also have no known geological explanation. Pixel noise and imaging artifacts can be readily ruled out by simple tests. These appeared in the raw data posted to the Internet worldwide within hours after receipt from the spacecraft, and share the noise levels and brightness and contrast ranges of the actual image. A few additional examples are seen at other locations in each of the three MGS strip images taken over the Cydonia region.
Figure 13. Extract from previous figure with standard noise filter applied to eliminate noisy or random background from image.