The water-powered clock tower with armillary sphere and celestial globe constructed in China in the Song dynasty was a precision astronomical instrument with an accuracy better than 100 seconds in 24 hours. Its effectiveness in tracking and observing heavenly bodies was many centuries ahead of Western instruments. Yet for the past 1000 years, no-one anywhere in the world--not even the British sinologist Joseph Needham--has succeeded in recreating it.
But three years ago, the problem was solved by a small group of researchers at the National Museum of Natural Science in Taichung. . . .
(In the Shaoxing period, 1131-1162, of the reign of the Song dynasty emperor Gaozong)
Su Xie was at home one day when his servant suddenly announced the arrival of the treacherous minister Qin Hui. Qin Hui had come to read Su Xie an edict from the Emperor, commanding him to build a "Yuanyou armillary sphere and celestial globe" for the court.
This was a precision apparatus used for astronomical observation and for timekeeping, which had been built by Su Xie's father Su Song in 1088, the third year of the Yuanyou period of the reign of the Northern Song emperor Zhezong. Su Song, who was Minister of Punishments and an expert in the calculation of calendars, led a team of skilled mathematicians and ingenious artisans to build the clock tower, which combined observational, demonstrational and time-keeping functions, and was the most advanced astronomical instrument of its day anywhere in the world.
The clock tower's main structure comprised three levels. The upper level bore an armillary sphere, used for astronomical observations; on the middle level was a celestial globe, which displayed the movements of celestial bodies; while on the bottom level were wooden mannikins which struck the time of day. The whole formed "a single mechanism, turned by the power of water, without human effort." Its precision was such that it was said to have "measured time as exactly as the sundial."
The armillary sphere on the top level comprised 12 rings in three layers, each of the rings being marked with a scale. From the armillary sphere one could directly read off the positions of the 24 Solar Terms, and by looking through a sighting tube one could find the star or planet one wished to observe, and define its coordinates. On the armillary sphere there was also a large gear wheel which connected it to the drive system of the entire clock tower. Thus the officers of the Ministry of Astronomy, which was charged with making astronomical observations, could use the sphere to follow the movements of the sun and stars.
But sadly, only 39 years after the clock tower was completed, the Jin army stormed the Song capital at Bianjing (now Kaifeng in modern Henan Province). They completely dismantled the clock tower and carried it off to their capital at Yanjing (now Beijing). But perhaps because the Jin could not work out how to reassemble it, the dismantled tower is said to have finally been left to gather dust.
Fortunately, after Su Song completed the clock tower, he wrote a book called Xin Yi Xiang Fa Yao ("New Design for an Armillary Sphere and Celestial Globe"), in which he described in text and diagrams the form, materials and dimensions of every component, and how to assemble the entire apparatus.
On receiving the imperial command, Su Xie hurried home to search out his father's old book, but after studying it long and hard, he was still at a loss as to how to set about the task. Su had heard that the great scholar Zhu Xi had an armillary sphere in his home and knew all about water-driven mechanisms, so he went to ask Zhu's advice. But he was still unable to successfully recreate the instrument.
The reason was that although Xin Yi Xiang Fa Yao contained over 60 complete views, part views and detail drawings of the clock tower, the only part covered more or less completely is the celestial globe--there are many gaps in the dimensions of the other parts. "Key sections of the book have been excised, probably to keep their content secret," was Zhu Xi's final judgement.
The lure of history
Mainland Chinese scholars observe that Su Song's astronomical clock tower is an important key to research into ancient astronomy and astronomical instruments. Since the invention by the Han-dynasty astronomer Zhang Heng (78-139 AD) of the world's first water-driven armillary sphere, most Chinese astronomical instruments were fitted with automatic drive mechanisms. But sadly, detailed descriptions of them have not survived--only brief mentions in astronomical annals from various periods. As for star charts, most from before the Song dynasty have also been lost. But Su Song's water-driven clock tower was a concrete embodiment of the achievements of the ancient Chinese in this field.
British scholar Joseph Needham wrote in an article published in the journal Nature in March 1956: ". . . the examination of certain medieval Chinese texts . . . has now permitted us to establish the existence of a long tradition of astronomical clock-making in China between the seventh and fourteenth centuries." He also wrote that escapements of a type similar to the one used by Su Song did not appear in Europe until the 17th century. Despite the precision and skill with which these instruments were crafted, the lack of detailed written records inevitably raises doubts in people's minds: if the Chinese really did have such advanced astronomical instruments and clocks a thousand years ago, why was the art of making them lost to people of later times?
To answer this mystery, researchers in mainland China and Japan, and British scholar Joseph Needham, have all attempted to recreate Su Song's clock tower, but without success: no-one has ever seen the mainland effort actually working; the piece displayed at the Tsukuba exhibition in Japan only recreated the main drive wheel and the timekeeping section; and although Joseph Needham early on made a model of the drive wheel, later for some reason he broke off from this project.
This "lure of history" was resurrected seven years ago when the National Museum of Natural Science was planning its Chinese Science Hall. The person "seduced" this time was the then museum director Han Pao-te.
China's "water culture"
The contract to design the museum's Chinese Science Hall had been won by the Japanese company Tanseisha. As Han Pao-te explains, at that time academic exchanges with mainland China had not yet been authorized, and although much research material exists on the history of Chinese science and technology, it was not accessible to researchers here in Taiwan. But the Japanese could get hold of mainland materials first hand. "Initially, Tanseisha wanted to put in a copy of the armillary sphere from the Qing-dynasty imperial observatory. But I didn't agree to this, because that sphere was made by foreigners, and I thought such an exhibit would be more decorative than scientifically meaningful," says Han.
Later, when Tanseisha presented a second plan proposing an exhibit based on Su Song's water-powered astronomical clock tower, which brings together advanced ancient Chinese technologies of astronomy, manufacture, mechanical engineering and construction, it immediately aroused Han Pao-te's interest and enthusiasm.
"Whether from the perspective of science, history or national culture, recreating the astronomical clock tower was very worthwhile." Han Pao-te, who now heads the preparatory office to set up the Tainan National College of the Arts, says that the Chinese were skilled in the use of water, which was the most common source of motive power. An example is the water wheels used to raise water for irrigation. "The astronomical clock tower represents Chinese water culture, and from the celestial globe and armillary sphere we can see the Chinese conception of the heavens, so they are an expression of the Chinese soul."
However, because at that time nobody had successfully reproduced the clock tower in its entirety, Tanseisha planned to contract out the job of making it to the famous clock and watch manufacturers Seiko, who were to base the timekeeping section around modern electronics--at an asking price of over NT$100 million.
Han Pao-te was well aware that recreating ancient science and technology is no easy matter, and with the opening date for the Chinese Science Hall fast approaching, to be sure of not ending up with an empty space when the time came, he saw no option but to agree. But to his surprise, after the proposal was passed up to the Ministry of Education for approval, the science historians called in by the ministry's science advisory office didn't support the idea. The reasons were firstly that it was too expensive, and secondly, that the Japanese had not previously succeeded in building a replica either.
However, at that point the construction company which was building the hall ran into financial difficulties. Construction would not be completed on time, so the opening seemed certain to be delayed. But this also meant that the project to recreate the astronomical clock tower was no longer under such pressure of time. "I changed my mind--rather than getting the Japanese to do it, which after all that criticism from the academics would not have seemed much of an achievement, why not take a crack at doing it ourselves?" says Han Pao-te feistily.
Puzzling over ancient texts
In fact, the only trump up Han Pao-te's sleeve was Kuo Mei-fang, a former student of his at NTU's Graduate Institute of Building and Planning, who by then was an assistant researcher at the Museum of Natural Science. At the museum Kuo Mei-fang had been involved throughout the process of preparing the project specification for Seiko, and at that time she was probably the person in Taiwan with the deepest understanding of the clock tower. Han Pao-te assigned engineer Wang Yung-hsin of the museum's exhibitions department to assist her in executing the project, and detailed two carpenters to make models for them. Thus this little group of people embarked on the project for which the Japanese had wanted NT$100 million.
For Wang Yung-hsin, who is quite a computer expert, making a model would not be difficult; what would be difficult would be to recreate the instrument as it really was in ancient times.
"To use modern methods to solve the problems involved would really have been very easy," says Wang Yung-hsin. He recalls that when museum director Han Pao-te first asked him to take part in the replica project, he slapped himself on the chest and said it would be no problem: for instance, to regulate the flow of water he could simply install a microprocessor-controlled valve, which could be made as accurate as it needed to be. But to his surprise Han Pao-te cut him off with a wave of his hand and said, "Absolutely not--we want to recreate the original, so we have to use the methods they used in ancient times."
To recreate the original, it was first necessary to get a clear understanding of the original design. "The hardest part was when we started reading Xin Yi Xiang Fa Yao," recalls Wang Yung-hsin with a grimace. "I recognized every character, but putting them together I couldn't make head or tail of it."
In his estimation, judging from the way Su Song refers to himself in Xin Yi Xiang Fa Yao as "your loyal minister Song," this book was a "report" written by Su for the emperor after completing the project. "As it was written by an expert for a non-expert, there would be no point in putting in too much detail." But how then were they to recreate it? "We had to infer the 'unknown' from the 'known,'" says Wang Yung-hsin.
For instance, for the mechanical part, Su Song's book mainly contains simplified sketches, not real working drawings. After Kuo Mei-fang worked out from these two-dimensional diagrams the three-dimensional structure and how the components fitted together, she drew out sketches showing all the dimensions, and gave these to part-time worker Hsieh Chih-hung, a student at Tunghai University's Graduate School of Architecture, to prepare computer drawings from them.
"Since these were only sketches, the dimensions could not be completely accurate, but an error of just a few millimeters is enough for the gears not to mesh properly," says Kuo Mei-fang. For this part of the work, modern computers were a tremendous help. However, in the final stage of actually making the parts, she enlisted the help of Professor Chen Tieh-cheng of the mechanical engineering department at National Cheng Kung University. Kuo Mei-fang printed out the size and rotational speeds of the gears, and from this Chen Tieh-cheng calculated the number of teeth they should have.
A tall order for a lowly civil servant
For Kuo Mei-fang, who had taken classes in Chinese architecture while in graduate school, had studied classical Chinese, and was familiar with ancient architectural drawings, reading Xin Yi Xiang Fa Yao was not so difficult. By studying it again and again she came to appreciate the tremendous ingenuity of many aspects of Su Song's original design, and solved many riddles (see "How the Astronomical Clock Tower Worked"). But working on her own, Kuo Mei-fang could not quickly find the answers to crucial questions: Just what did the scoops on the drive wheel look like? How did the "upper locks" work? Fortunately Wu Ming-hsueh, a part-time volunteer at the museum, also worked at the Center for Aviation and Space Technology and had much experience in the construction of mechanical components and the properties of materials. He helped Kuo Mei-fang pick her way through the maze of information. Furthermore, when Kuo went to England to attend an academic conference in honor of Joseph Needham's 90th birthday, she got to know a mainland Chinese science historian, and with his help began to receive large quantities of information from the mainland.
Thus she began to read mainland research papers and to compare them carefully with Xin Yi Xiang Fa Yao: How did they differ? What effect would the differences have? Why not do things such-and-such a way? In this way, step by step she constructed an initial plan, and the key points gradually became clear. "I was definitely building on others' work," she says.
Interpreting the design drawings was no easy matter, but even when she had understood the principles of the clock tower's operation, it did not mean all the problems had been solved. Taking materials as an example, the capacity and weight of each scoop had to be the same to ensure that each accepted the same amount of water. But using natural materials such as bamboo and wood, how could one find pieces of exactly the same thickness or grain density?
This was the most bothersome problem for Kuo Mei-fang. "I was sure the clock tower could be made to be very precise, but only if it was built by very skilled craftsmen. This might have been possible for an ancient emperor to arrange, but for me as a lowly civil servant it was a tall order."
A new ancient monument
But fortunately, whatever requests she made, her two bosses at the Museum of Natural Science--director Han Pao-te and exhibitions department head Kao Chen-hua--did their best to support her, whether in meetings or with approval processes. "They gave me the greatest flexibility possible within the civil service structure." For this, Kuo Mei-fang is still full of gratitude. And on the technical side, Han Pao-te's many years of previous experience in the restoration of ancient monuments, and the contacts he had built up, were a great help too.
Today what impresses museum visitors most is the air of antiquity imparted by the clock tower's wooden structure, and the wooden mannikins behind the windows in the five levels of its pagoda-like facade, which mark the time of day by striking bells and drums. The bells mark the chen--the 12 two-hour periods of the day, while the drums mark the ke, which divide the day into 100 equal periods of 14 minutes and 24 seconds each.
This part of the project was the responsibility of the Ching Jen Construction Company, which has over 20 years' experience in the restoration of ancient monuments. Xin Yi Xiang Fa Yao contains only one general sketch of this section, so when Kuo Mei-fang entrusted the work to Ching Jen owner Lin Jui-hsiung, apart from drawing out the shapes of the dougong (supporting brackets) on each level and marking the main dimensions, she left the other details up to him.
"This was the most challenging job we'd ever taken on. It was very, very interesting." Three years on, Lin Jui-hsiung, who has kept all the design documents from the time, explains that because space inside the heavy wooden structure was limited, it was necessary to plan from the start how to maneuver each wooden section into position--any mistake might mean having to take the whole thing apart and start again. Furthermore, ancient Chinese carpenters did not use metal nails, so Lin had to decide what kind of mortise and tenon was appropriate for each joint. There was no way this would be illustrated in the book, so he had to rely entirely on his experience and skill.
But however hard they worked, because of the enormous span of time between the Song dynasty and the present day, to create a 100% replica was not only impossible, but perhaps not even necessary.
For instance, to find clues as to what the mannikins which mark the time of day might have looked like, Kuo Mei-fang consulted all kinds of images from that age, from Song-dynasty molded tiles to illustrations of theatrical performers. On one picture she wrote the following comments: "This can serve as a model for the sculptor. Perhaps these brothers are twins. They seem partial to fat meat--they're a bit too stout. Can we ask the sculptor to slim them down a bit, and adjust the proportions of the head and body--the legs are too short! Only then will these fellows be suitable for the job of time-keeper." Later, Lin Jui-hsiung really did make the figures' faces thinner and the legs longer, and this is how the clock tower's wooden mannikins were designed.
Ancient forms, modern materials
It was not possible either to completely recreate the materials of the original clock tower.
According to the astronomical annals in the History of the Song, after the clock tower was completed in wood, it was copied in bronze, using a total of "20,000 catties" (around 12,000 kg) of the metal. But when building the replica, the Museum of Natural Science faced the problems of finance, time and technical feasibility. Firstly, bronze is expensive, and funds were not sufficient to remake the tower in the metal. Secondly, to do so would have taken at least two years, and the museum could not wait. But most seriously, bronze is too heavy. Just the equatorial ring of the celestial globe would have weighed over 200 kg. If the armillary sphere and celestial globe were made entirely of bronze, "we would need a crane to get them up there. This would have meant we would have had to first construct the clock tower, then build the museum around it," says sculptor Lin Chien-cheng, who took on the job of making the sphere and globe.
But the museum was already built, and they certainly couldn't tear it down and build it again. So what could be used instead of bronze? The material most often used to simulate bronze is fiber-reinforced plastic with a coat of metallic paint. But this would not be realistic enough, and after some time the paint would begin to flake off. For better effect, Lin Chien-cheng added powdered metal to an acid- and alkali-resistant, non-oxidizing resin which he formed manually into the shapes required. He then polished them to create an effect which really does look completely like bronze.
In fact, Kuo Mei-fang believes that if they had wanted to remake the tower in bronze, it would not have been possible without the assistance of a bronze foundry. "The various different components all require differently alloyed bronzes. For instance, the metal for the scoops has to be corrosion-resistant, whereas the gears have to be hard-wearing. Bronze smelting skills in ancient China were very advanced. We would need a bronze research institute to achieve the same results today."
Red and black stand in for five colors
Although the tower could not be recreated in the materials of the Song dynasty, Han Pao-te, who is himself an expert in Chinese architecture, was able to provide much advice where form and design were concerned. For instance, how many digits should the dragons' claws have? And how should they be curved to match the Song style?
Kuo Mei-fang had the positions of the constellations on the celestial globe checked by Professor Huang I-nung of Tsinghua University's Institute of History, who has a doctorate in astronomy. This threw up another difficulty: according to Xin Yi Xiang Fa Yao, the constellations on the globe were marked out in "pearls of five colors." But ancient books were not printed in color, so one can only see black and white. Since it was impossible to determine which stars were represented by which colors, the best they could do was to use artificial agate in red and black to stand in for them.
There's another feature of the clock tower which is different from the Song dynasty original: it's missing its roof. This is because the Song dynasty tower stood out of doors, whereas the one at the Museum of Natural Science stands inside a large hall. So why didn't they decide in the first place to build this one outside too, to allow visitors to climb up on the tower themselves and use the armillary sphere to observe the stars and compare them with the celestial globe on the middle layer? Wouldn't this be a still greater testament to the wisdom of the ancients? Or if this precision instrument really had to be built indoors to protect it from the elements, then couldn't they have put a glass roof over it?
Unfortunately this kind of bright idea benefits too much from hindsight, because the museum building was erected before the clock tower was built. Another factor is that the Song capital lay at a different latitude from Taiwan, so the constellations seen here do not match those depicted on the celestial globe.
A further area where ancient methods were discarded in favor of modern ones was the gear wheels, because "we had to rely on machines--we couldn't have the gears cut to the original shape by hand," says Kuo Mei-fang. Fortunately all but one of the main gears could be cut to the correct number of teeth using modern machines, but they do not run with complete accuracy over long periods.
Finally Kuo Mei-fang came to a deep realization: "I'm sure the water-driven clock tower really did exist and was not merely a fabrication on Su Song's part, otherwise the dimensions he gives in his book could not have been so accurate. Also we had to make the pieces exactly to his dimensions, otherwise the various components wouldn't have fit together properly."
Finally it was time to assemble everything. At this point everyone was full of anxiety as to whether the contraption would actually work, and if it did work, whether it would be accurate. Once it was fully assembled, it really did complete one revolution per day, and everyone forgot their worries.
However, time really had been too short, and when the Chinese Science Hall opened, the clock tower on display was still the development prototype, not the final finished version, so it didn't run very smoothly. For instance, when wood gets wet it swells, and the drive wheel often jammed, or else the escapement would disengage and suddenly skip several scoops, spilling water all over the floor.
Improvements have been made continuously, and this June the water raising system will be changed to different materials. The present wood and bamboo parts will be exchanged for aluminum alloy anodized in a bronze color, to ensure a constant water supply; apart from this, for lack of manpower the museum "couldn't possibly do as they did in the ancient imperial palace and have somebody fetching water all day," so the water which runs out of the scoops is recycled directly back to the water supply system, and not allowed to run to waste as in Su Song's design.
For greater authenticity, this June a clutch will be installed too: normally the water is pumped back up to the top, but during performances, museum staff will be able to disengage the motor and let the public carry water themselves, to get a feel for how this small astronomical observatory really operated in Song times.
Completing the world's first full-size, dimensionally correct working replica of Su Song's water-powered astronomical clock tower, which combines many different technologies, brought the National Museum of Natural Science to international attention. Lin Chien-cheng, who made the celestial globe and armillary sphere, has even been sought out by Japanese museums to work with them.
Just a big cuckoo clock?
The reaction of the public in Taiwan has also been very enthusiastic. During the Chinese New Year holiday early this year, as part of its activities to mark the 10th anniversary of its opening, the Museum of Natural Science asked the public to vote for their favorite exhibit in the Chinese Science Hall, and the clock tower came out in first place. But Wang Yung-hsin, who often carries out maintenance work on exhibits in the hall, admits that most visitors do not look at the tower in much detail. They stand at the bottom of tower, look at it for a minute, take a picture and then move on. "At most they see it as a work of art, but they are not interested in really understanding its significance."
But what exactly is the purpose of exhibiting this kind of ancient Chinese technology? Most people only appreciate it on a superficial level, admiring its ingenious form and impressively heavy structure. If people get no more from it than from watching the cuckoo popping out of a cuckoo clock, it rather seems to negate the purpose of recreating this ancient astronomical observatory.
This is just the point on which scholar Huang I-nung is not entirely satisfied: the museum should not simply recreate the clock tower itself, but should also explain what problems were met with during the recreation process, how the researchers solved them, and how the ancient astronomers used this tower for their work. "If they could make an audio-visual slide show to let visitors learn about these aspects as well as about the structure itself, people would gain much more." In Huang I-nung's view a successful museum must satisfy the needs of all different age groups: elementary school children must be able to enjoy themselves, while senior high school students should be able to satisfy their great thirst for knowledge.
Of course, if perfection is the goal there is still much room for improvement. But compared with many other national grade museums being built or being planned in the ROC--such as the National Science and Technology Museum in Kaohsiung, which is also to have a Chinese science hall, but which has contracted all the exhibits to a Japanese design company--the National Museum of Science's successful and innovative creation is definitely praiseworthy.
In the words of an old poem, "I love the ancient melodies, but few play them today": the astronomical clock tower stands out as a rare example in Taiwan today of the successful recreation of the products of ancient Chinese science and technology.