Sundial, calendar and Khmer temples

Astro-archaeology = archaeo-astronomy = Astroarchaeology = archaeoastronomy

Prasat Phu Phek

1711.5296' N - 10356.2356' E

The northernmost ancient Khmer 'mountain-temple'


Above: Phu Phek Mountain in the left side of the picture. The picture was taken from a position straight true east of Phu Phek around summer solstice 2004. The sun is setting at its northernmost position and will set at the peak of Phu Phek at equinox.
Content of this web-page:

1. Introduction
2. Dating Prasat Phu Phek
3. An old myth about Prasat Phu Phek
4. A new myth about Prasat Phu Phek
5. Determination of orientation
6. The square pedestal as a calendric devise.
7. The equinoctial point


1. Introduction

     Prasat Phu Phek ('the temple on the Venus Mountain') is located 22 km east - 'as the crows fly' - of Sakon Nakon City and is the northernmost 'mountain-temple' in the Ancient Khmer power-sphere. The term 'mountain-temple' is in this text defined as a temple on a mountain-top. The term 'temple-mountain' is widely used in the literature as a temple constructed on an artificial 'mountain' or earth hillock, ''where the principal idol was placed on top of a stepped pyramid'' (Briggs, p. 202). Both kind of temples share the same feature as being elevated above the ground to a level where the horizon is visible, which makes the site a well qualified place for astronomical observations. With it's elevation of app. 520 m above sea level the temple is the highest located prasat in the whole of the ancient Khmer Empire. At the same time Prasat Phu Phek also indicates the northernmost extent of the Khmer power sphere in the 11th century A.D.
     The author of this paper has conducted studies and experiments on Prasat Phu Phek since his first visit in 1999 - partly because of its unique elevated location, partly because Prasat Phu Phek is only 70 km from his astronomical test-site, where he has a gnomon-style sundial and ongoing experiments in constructing solar-gates true east-west only by the use of the sun  in an attempt to find out how ancient peoples could have determined equinox day and the equinoctial point, where the sun rises on equinox day.

     Sakon Nakon province, which is located in the north-eastern corner of NE-Thailand (Isan), hosts at least 3 other ancient Khmer sites: Prasat Naria Cheang Weng (Baphuon to Angkor Wat style), Prasat Ku Phanna (a Jayavarman VII 'hospital' chapel in Bayon style) and the 'bridge' Saphan Khom (probably a Jayavarman VII 'ceremonial bridge' in Bayon style). There might be more sites as some present day Buddhist temples might have been constructed on ancient Khmer sites.
     North of these sites we only know about one more Khmer temple: A Jayavarman VII temple at Vientiane in Laos P.D.R.

Above: A GPS-generated map showing the location of Prasat Phu Phek. The temple is located straight east of Nong Han Reservoir and nearly straight east of Prasat Chaeng Waeng (2 degrees of).


2. Dating Prasat Phu Phek

     We have no written records as f. ex. engraved steles in Sanskrit or ancient Khmer about Prasat Phu Phek. An estimation of the age can only be done indirectly by comparing  the style of art between this temples and similar temples, which contains written records.

     The Fine Arts Department of Thailand dates in the book, Ancient Sites in NE-Thailand, the site to ''probably belonging to the Baphuon art style'' (1050 - 1080 A.D. covering the reigns of Udayadityavarman II and Harsavarman III). One signpost on site gives the same dating. A second dates it to the 11th - 12th century A.D. and thereby including Angkor Wat style. The latter signpost continues: ''This is a Hindu sanctuary comprising of a large east-facing sandstone tower and two small ponds in the north'' . . . ''As no fragments of the superstructure were discovered, it is assumed that the tower was incomplete. Two pieces of important objects were discovered inside the sanctuary; namely, a stone base with a square hole in the middle believed to have once held auspicious items and the square base and octagonal middle part of a Siva Linga that is usually topped with a cylindrical part, but broken of.''
     A signpost on the way to Sakon Nakhon (Thai Tourism Authority?) dates the temple to the reign of Jayavarman VII (1181 - ca. 1220 A.D.) without arguing why. A local Thai doing astro-archaeological research on Prasat Phu Phek, Khun Sansonthi, also favours the fervent Mahayana Buddhist emperor Jayavarman VII as the initiator of the temple. He also argues that the the linga is erected by local followers of Siva (sic.).

     The author of this paper believes that Prasat Phu Phek dates to Udayadityavarman II (Baphuon, 1050 - 1066) based on the 'pyramid shaped' base stones, which can also be seen at other definitely Baphuon style structures. The stone base or pedestal and the Siva lingam are widely found at other Baphuon to Angkor Wat temples.


Above: Eastern facade Prasat Phu Phek
Above: Prasat Phu Phek

Above: Plan of Prasat Phu Phek (courtesy to F.A.D.).
Above: Central chamber       Above: Original setting.


3. An old myth about Prasat Phu Phek

     ''A local ruler, Suwannapikka, who built the temple for veneration of Lord Buddha's ashes and Phra Mahakasapa, who was second to Lord Buddha, used to come to worship at Prasat Phu Phek together with 500 other monks. So the ruler decided to build two prasats! One on the top of the Phu Phek (phu = mountain, phek = Venus) Mountain and another, Prasat Naria Cheang Weng, on the plain. The first temple had a male work force and the second a female. The two groups should compete on the construction and the winner would host the ashes of the Lord Buddha. They were both told that the construction must be completed before Venus became visible at the horizon again. The female group faked Venus by the use of a balloon with fire hanging below in order to make the men believe it was Venus and when the men saw the fire-ball, they thought it was Venus and stopped working. And that is why the temple has been left unfinished until today.''

     This is how the story was told to the author some years ago. Later on he became aware of that the story or myth is not that local. The French explorer Etienne Aymonier, who among other tasks, conducted survey on Khmer temples in SE-Asia in the last part of the 19th century, reports that he heard variations of the same story many places in Cambodia, Southern Laos and NE Thailand.

     Notes on the myth:
     The legend is about male and female powers, who are competing with one another - with a celestial dead-line: The rise of Venus.
     Venus is invisible for the human eye, when it is 'too close' to the sun. When the angular separation from the sun becomes sufficient Venus becomes visible. This is also called heliarcal rising and has been described in various Hindu astronomical text-books, Siddhantas, many hundred years before the construction of the two temples concerned in the story.
     The names Suwannapikka and Phra Mahakasapa are neither ancient Khmer nor Sanskrit - and the two temples are Hindu, not Buddhist.

Right and left:
An ancient Buddha figure now located at the Wat Phu Phek temple at the foot of the mountain. Post Bayon?


4. A new myth about Prasat Phu Phek: Phu Phek as a 'solar calendar'

     Visiting Phu Phek around winter solstice 2002 the author noticed a hand-written sign in front of the pedestal right outside the central sanctuary saying Surya Phatthithin, 'solar calendar', and returning to Sakon Nakon at summer solstice 2004 the above mentioned Thai Tourism sign-board gave the same information, which now apparently had been upgraded to official truth.
     Searching for data and arguments the author visited The Cultural Centre of Sakon Nakon at Sakon Nakon Rajabhat University, where the lecturers knew nothing about the matter. Later the same day a scholar at the language faculty brought a report from the tourism faculty - a report written in Thai by Mr. Sansonthi.

     The report can not be regarded as a scholarly work as it contains no data which can be checked or verified. The conclusion is that Prasat Phu Phek is a 'solar calendar', Surya Phatthithin, with analogies to Stonehenge and Mayan temples. The calendar allegedly tells the days of solstice and equinox. One central argument is that the holes in the square pedestal (wrongly referred to as a yoni) are aligned with the equinoxes and solstices. One 'proof' hereof is that the pedestal should be oriented towards the equinoctial point (sunrise on equinox day). One picture shows a compass indicating that the orientation is 90 degrees magnetic east. Another 'proof' is a photo showing a person holding a plump in a plump line with the rising sun behind, and apparently taken some 10-15 minutes after sunrise on solstice day. A third 'proof' is photos showing the rising sun above the broken linga on equinox day, also some 10-15 minutes after sunrise. The technical arguments is the use of an astronomical computers programme, but there is no mention of data-in-put or data-out-put.

     Comments on the photos:
     1. The first photo showing that the pedestal is oriented 90.0 degrees from magnetic north, indicates that the true orientation is 87.5 and thereby not pointing at the equinoctial point. Magnetic deviation is not taken into account, which is a serious mistake.
     2. The second photo showing a person holding a plump has several errors: One is that there must be two plump lines to depict an alignment (just like we in basic geometry need two points in order to construct a line). The second error is that the sun some 10-15 minutes after sunrise has changed azimuth (see red line on the photo below). The third error is that a sightline on only 50 cm cannot give accurate readings because the position of the rising sun around the solstices barely changes from day to day - a sightline should be in the range of 10 km. And fourthly: The object in question, the pedestal, is not oriented straight true east (see '1' above). And lastly: The object in question is a square, where the angle in question is close to - but not identical to - the solstitial angle (for details see ''6. The square pedestal as a calendric devise'' below).
     3. The photos below, showing the rising sun above (the broken) linga on equinox day, proofs nothing. Both photos are taken after sunrise and not from the same place. The photo to the left is not taken from the centre-line, which can be seen from the uneven door-frames. The photographer has apparently moved to a position so the sun is visible on top of the linga (ad hoc!). The photo to the right seems to have been taken close to the centre-line of the structure. The author of this paper has added the red line depicting the path of the rising sun and the white circle depicting the assumed equinoctial point (for a more accurate photo analysis, see ''7. The equinoctial point'' below).

Photo above: Vernal equinox 2000.
Courtesy to Mr. Sansonthi.
Photo above: Vernal equinox 2000.
Courtesy to Mr. Sansonthi.

     Prasat Phu Phek can therefore not be regarded as a 'solar calendar' telling the equinox days and the solstice days. Especially not the pedestal and the linga, which have been re-located from their original or intended place. Neither of these two objects are parallel to one another, nor are they oriented towards the equinoctial point, nor are they levelled.
     Prasat Phu Phek could to a certain degree be regarded as a 'solar calendar' if the centre-line of the temple is proven to be oriented straight true east. If so, then the rays of the rising on equinox day (Vedic New Year) would illuminate the linga (standing in the central sanctuary) entering the temple through the unfinished eastern doorway(s).

     The author believes that this was the intension of the constructor of Prasat Phu Phek (to visually depict Vedic New Year). But this does not make Prasat Phu Phek unique: Approximately 30% of all ancient Khmer temples, barays (water reservoirs) and ancient settlements are oriented true east.


5. Determination of orientation

     This chapter will demonstrate several methods used by the author when determining the true orientation of a man-made construction.
     All methods are used by the author in his since February 2004 ongoing field-research aimed at determining the orientation of all ancient Khmer structures in NE-Thailand and Southern Laos. Later on some of the nearly 3000 temples in Cambodia will be visited.

5.1. Solar noon
     The idea in this method is to observe the shadows of the sun at solar noon, when the sun passes the meridian - this is also called the culmination of the sun. If the structure in question is oriented straight true north-south then the shadow will be aligned with the structure at solar noon.
     This method is not 100% exactly. Primarily because it is difficult the decide exactly when the sun is aligned with the structure. Secondarily because the structure has to be vertical, which is seldom the case with an ancient structure - and not either on a renovated structure. The method only gives a rough hint and has to be followed up by more refined methods.

     The photos below show the shadow in the eastern doorway of Prasat Phu Phek. It was calculated that solar noon on equinox day, the 21st of March 2002, would be at 12:11:31. Photo no. 3 below (at 12:10:00) was taken the shadow apparently was aligned with the doorstep. The difference between calculation and observation is 1:31 minute and indicates that the doorway is either not oriented straight north-south or that the top of the doorframe is leaning westwards.
     From the author's field-notes: ''Follow up - this structure is close to having equinoctial orientation!''

12:04:00 12:06:00 12:10:00 12:14:00


5.2. Compass

     Using a compass to determine the orientation of a structure gives more accurate results than method 5.1., even some obstacles are involved. One is magnetic deviation, which is the difference between magnetic north and true north. Magnetic deviation is different on different locations. Another obstacle is to decide fix-points to align.
     The equinoctial point is related to true north as it is located true east. A military map will inform about local magnetic deviation. The author prefers to determine magnetic deviation himself. At his test-site in the Phu Phan Mountain Range some 70 km from Phu Phek the local magnetic deviation is determined to be 2.5 and this figure is used at Phu Phek as well. At Prasat Phanom Rung the gateways are determined by solar observations to be oriented 84.5 and as the compass shows 87.0 the same figure is used there too - and in the area between too.
     The difficulty about deciding fix-points to align is not existing when measuring for example a straight vertical wall. But measuring a ruin can often be difficult. The best solution is to find (dig!) after foundations and then raise vertical bamboo sticks, but this is illegal.

     Using compass there are two choices: A military compass or/and a map-compass. The first for long alignments the latter for short orientations.

Above: Military compass
Above: Map Compass

Note that the compass above is not parallel to the green string. The string is oriented true east aligned by the sun. The compass is aligned 90 magnetic east. The difference is magnetic deviation.


Compass readings at Prasat Phu Phek:

     Military compass:
     The temple along the centre-line from the eastern staircase towards the central sanctuary = 182.5
magnetic west => 92.5 magnetic east. And 92.5 magnetic east minus 2.5 magnetic deviation = 90.0 true east.

     Map compass:
The pedestal: 90.08 magnetic east => 87.58 true east.
     The lingam (averaged): 87.0
8 magnetic east => 84.58
true east.


5.3. GPS measurements

     A GPS device can be used to determine orientation too. Two methods are used. One is calculations based on waypoints, but due to the accuracy on 6 meter (decided by the US military) long distances between the waypoints are needed in order to obtain reasonably accurate results. Another method is 'track-measurements' by which one can obtain good results (and better than compass readings) on distances from 50 m and more. Both methods are described elsewhere and were not used at Prasat Phu Phek due to the short sightlines.


5.4. Sunrise observations

     Observations of the rising sun aligned with the centreline of the temple is the ultimate and most accurate method. This has not been done at Phu Phek because the author at equinox is checking his solar-gates on his test-site for experimental astronomy.
     In order to conduct sunrise observations at Prasat Phu Phek an interim eastern gate at the eastern staircase should be raised. At least two gates are needed to determine the location of the rising sun in relation the the structure of the temple.


6. The square pedestal as a calendric devise.

     The pedestal at Prasat Phu Phek is among archaeologists referred to as a 'pedestal for placing auspicious objects'. The auspicious objects are supposed to have been placed in the small square holes. The central hole is thought to have hosted a religious figure. The pedestal must not be confused with a yoni, which hosted a linga and also has a drain for auspicious water to flow of. These pedestals are found at many different ancient Khmer temples and are either square or rectangular. The pedestal found at Prasat Phu Phek belongs to the square type.

     The pedestal at Prasat Phu Phek is by Mr. Sansonthi being regarded as a surya phattithin or a 'solar calendar'. The argument is that the alignment from the centre of the central hole on the western side of the pedestal to the centre of the corner hole(s) of the eastern side will point at the rising sun at solstice (see drawing below).
     But this is not correct: Any given square will from the mid of one side to the corner of the opposite side have an angle = 26.57. And the angle of the rising sun from the azimuth of the equinoctial point to the azimuth of sunrise at solstice is in 2004 A.D. = 24.60 at the latitude of Phu Phek (this angle is from here on called the solstitial angle). The difference is ca. 2 (see drawing and calculations below).

Above: Photo of the pedestal.

Right: Schematic drawing of the pedestal.


The solstitial angle at Phu Phek:    
At the solstices 2004 A.D. (the sun 50% visible):    
Winter: 114825'21''  
Summer: 65813'35''  
Difference: 49811'46''  
'Solstitial angle' = the half of the above difference: 24835'53''  = 24.608
At the solstices 1066 A.D. (the sun 50% visible):    
Winter: 114832'48''  
Summer: 65806'13''  
Difference: 49826'35''  
'Solstitial angle' = the half of the above difference: 24843'18''  = 24.728
The difference between the solstitial angle in 2004 A.D. and 1066 A.D.  =   0.128 (= 0807'12'')

8 might not sound like much, but compared with the diameter of the sun, which is 0.528 the difference comes out as 23% of the diameter of the sun (refraction not accounted for).
And when comparing
0807'12'' with the differences in azimuth of the rising sun in the days around solstice as brought below the difference becomes rather erroneous .


Azimuth of the rising sun at Phu Phek around winter solstice 2004:

The 21st of December: 114825'19''

The 22nd of December: 114825'21'' => difference from the day before = 0800'02'

The 23rd of December: 114824'52'' => difference from the day before = 0800'29''

The 24th of December: 114823'56'' => difference from the day before = 0800'56''

The 25th of December: 114822'25'' => difference from the day before = 0801'31''

The 26th of December: 114820.27'' => difference from the day before = 0801'52''

The 26th of December: 114817'59'' => difference from the day before = 0802'28''

     The small differences in the azimuth of the rising sun from day to day around the solstices clearly demonstrates how difficult it is to determine solstice day by solar observations unless the alignments are 10 km or more. On an alignment as the pedestal it is impossible.

Final remark: As the solstitial angle changes with time - as demonstrated above - it also changes by latitude: A more northern latitude will have a larger solstitial angle, so if the square pedestal was moved to somewhere in southern China, then the solstitial angle would be identical to the 'half-diagonal angle' on 26.57 - and the pedestal (if enlarged) could serve as a 'solar calendar', but the facts are that these pedestals are only found within the extend of the ancient Khmer Empire.

The pedestal and the linga are not aligned with one another.

Above: The compass reads 90.0
indicating the the true orientation is 87.5



7. The equinoctial point

     The photo to the right was kindly e-mailed by Mr. Sansonthi to the author in 2000, but lacked a forth-sight to give a certain orientation of the structure. It also lacked horizon, location of camera, time down to minute, and a clear shape of the sun. The only information was vernal equinox 21 March 2000 and the location.

     With the lacking information the author would be able to determine the equinoctial point, where the sun rose at mathematical horizon on equinox day.



      In an attempt to establish the alignment and the horizon the author visited Prasat Phu Phek a few days before winter solstice 2002 together with some of the workers who had been trained in levelling and aligning at the author's test-site. They also brought bamboo sticks for alignment and a long garden hose of clear plastic for levelling.

  Above: Levelling


Left: Aligning the centre-line of the structure from mid eastern doorway of the central sanctuary to mid eastern staircase.

Compass-reading: 92.5

Note that the pedestal and the linga are placed outside the centre-line and that they are not in level.


Right: Photo-analysis

     Based on enlargements of the photos above the author first drawn the mathematical horizon and the alignment on the above given photo. Then the path of the rising sun was added by drawing a line (blue) with an angle similar to the latitude of the location from the sun. The yellow circle indicates the assumed location of the rising sun, which should be very close to the equinoctial point, which again should be located where mathematical horizon intersects with the alignment line. But it is not so.
     The major reason is that equinox day in 2000 was the 20th - not the 21st of March (!).


Right: Sunrise at Phu Phek 21 March 2000

     The analogy between the 2 depictions of the rising sun to the right and the photo of the rising sun above is clear. This indicates that the equinoctial point is located where the mathematical horizon intersects with the alignment line - and that the structure is aligned towards the equinoctial point.

      The author looks forward to take a photo of the rising sun from the centre-line of the rising sun around equinox 2005 and at that occasion also determine the actual local horizon, which is lower than mathematical horizon due to the elevated location of Prasat Phu Phek


The Ancient Khmer Empire, Lawrence Palmer Briggs, Philadelphia, 1951.
Ancient Sites in NE-Thailand. Fine Arts Department (F.A.D.). Bangkok.
Khmer Heritage In Thailand, with special emphasis on temples, inscriptions and etymology, Etienne Aymonier, Bangkok, 1999.



12 October 2004