Here we have a volume devoted to a demonstration of the calculations of the Chinese calendar. The practical day-to-day calendar continued for over two millennia, and in all that time we have a regular pattern of years of 12 or 13 months, months which run from New Moon to New Moon, of 29 or 30 days. In fact this results from an ever changing set of rules, rules which are expressed in the astronomical chapters (律 曆 Lü Li) of the 24 Histories, and so available to any historian who cares to take up the matter. These rules express such astronomical theory as the Chinese ever developed. Subject to constant revision, there were in fact many distinct sets of procedures, or ‘calendars’; Martzloff has also provided a convenient list in an appendix to the Grand Dictionnaire Ricci. In the Brief History of Chinese Astronomy (中 国 天 文 学 简 史, Tianjin 1979) as many as 104 rules are listed.
These astronomical chapters are concerned with rather more than the calculation of the Sun and Moon (which includes both mean and true longitudes), but also extend to include the lunar node (required for the calculation of eclipses), the sidereal year, and the planets. In the present study however only the Sun and Moon are considered. The rather restricted aim here is to show how the practical details of the published calendar are derived from the rules of calculation of the Sun and Moon, while ignoring any matter concerning eclipses or the planets.
The volume opens with an account of the matters essential to a basic grasp of the published calendar, including the simple technical details, such as the system of numerical notation in the 60-fold cycle, expressed by the 10-fold stem and the 12-fold branch, and the 24 qi of each year, defined essentially by 15° steps in the Sun's longitude. It is explained how the collection of defining parameters, here referred to as C.A.O. (canons astronomiques officiels), prepared by the team of astronomical specialists, were subject to constant reform as the result of regular observations, such as the time of the Winter Solstice. Martzloff has here (pp. 45–53) interesting remarks on the Chinese reaction to disagreements between calculation and observation. The preparation of these canons was the work of the Bureau of Astronomy, an institution not unlike modern agencies in Greenwich, Paris and Washington. Here the difference from the modern institution, as I see it, is that the fundamental astronomical theory employed in the official western bureau is the creation of theoretical astronomers who stand quite apart in an academic world.
The response of Europeans to the Chinese calendrical practices has on the whole been deficient. Joseph Needham in particular was quite disappointing, and he remarked that “The whole history of calendar-making is therefore that of successive attempts to reconcile the irreconcilable, and the numberless systems of intercalated months, and the like, are thus of minor scientific interest.”; Science and Civilisation in China, Vol. 3, 1957, p. 390. At the time of Needham's preparation of that volume, 1957, he was visited by Kiyosi Yabuuti's disciple Shigeru Nakayama who attempted to persuade him that Yabuuti's profound research into the Chinese calendars (Investigation of Calendars of the Sui and Tang, 隋 唐 曆 法 史 の 研 究, Kyoto 1944) showed how Chinese theoretical astronomy, as far as it went, was revealed in these calendrical procedures; http://homepage3.nifty.com/shigeru-histsci/work3needham.html.
After this preliminary account of the external features of the public calendar Martzloff proceeds to explain the underlying mathematics. Briefly, the opening statement of the canon, as defined in an astronomical chapter of a history, will declare three steps that determine the character of the particular calendar. To begin with we are told the date of inception, specified by the year of a certain Nianhao, with a declaration of the number of years (A) elapsed since the Great Origin (Shangyuan). This is, to be precise, the number of years up to the Winter Solstice just preceding the date of inception. There is then a statement of the lengths of the year (Y) and month (M), each expressed as a ratio of integers. The number of days from the Shangyuan to the Winter Solstice occurring just prior to the date of inception is AY, and this defines much of what is needed to fix the ‘coordinate system’ of this particular calendar. Thus, if multiples of 60 are removed from AY to give a residue a, (AY≡a (mod 60)), then a is the sexagesimal number of the day of the Winter Solstice. If multiples of the length of the month M are removed from AY to give a residue b, (AY≡b (mod M)), then b is the interval to the Winter Solstice from the first day of its month (which is normally counted as the 11th month). This number of days AY can be converted in western terms to give the Julian Date of the Shangyuan. For example, in the Xuanming li (defined in Xintang Shu j.30A) we have “The 1st day of [Tang Dynasty] Changqing year 2 (root 9, branch 3) [39 in the 60-cycle] [whose first month starts on 822 Jan 27] is 7070138 years from the origin at (1, 1)[1]”. It is a relatively straightforward matter then to determine that the Julian Date of the Shangyuan is ‑2580308748.5, where the ‘.5’ indicates that this is midnight. This remote hypothetical midnight is not only a Winter Solstice, but a moment when all the mean longitudes coincide with the Sun, forming one grand conjunction; this includes the mean Sun, mean Moon, apogee and node of the Moon, and the five planets.
In Martzloff's volume we have detailed examples of the derivation of the calendar according to four canons, Sifen li (AD 85), Jingchu li (AD 237), Xuanming li (AD 822), Datong li (AD 1368). In the 3rd and 4th of these there are corrections for the ‘equations’ of Sun and Moon, corrections which are added to the mean longitudes in order to allow for their non-uniform motions. It is in explaining these equations that the clarity of the exposition here leaves something to be desired.
In the calendrical chapters of the Chinese text the equations of the Sun and Moon are defined by tables, in which the rows correspond, respectively, to the 24 qi of the year, or to the 28 days of the anomalistic month. Although not immediately apparent, each table provides essentially one quantity, a function roughly equivalent to a sine-shaped curve, the equation of the Sun or Moon. However, in order to assist in interpolation, and for the easy application of the equation, the table lists not only the successive differences of this quantity, but also a duplication of these two columns, differing from the former only by a constant numerical factor. Thus there are four columns altogether, although in reality one of these would suffice, since the other three could be derived by simple operations. There are also indications of plus/minus prefixed to the entries, but making use of a variety characters in somewhat unexpected and possibly confusing ways.
The example given here employs the solar and lunar equations of the Xuanming li; pp. 179, 190–194. The two columns of differences are denoted respectively δ, Δ, but without drawing attention to the fixed numerical ratio between them. The basic aim is to discover the effect of the equation on the precise time of a given lunation. Since the lunation occurs on some day lying in the 15 day interval between two successive qi it is necessary to interpolate, and the Chinese devoted much effort into improvements on simple linear interpolation. For they were clearly persuaded that linear interpolation was inadequate, and they sought to improve this by means of a variety of procedures leading to a quadratic contribution, corresponding to something like the first part of the Newton forward difference formula. Moreover it is the column Δ, and its summation (C = Σi Δi), that were employed to determine the correction of the time of the given lunation; of course the numerical factor between δ and Δ has been set to ensure the correct result. The column δ was used to make an adjustment to the interval between successive qi.
The resulting interpolation formula, which depends on both δ, Δ is expressed as a formula given on pp. 190–191, eqns (5.18–5.22). Here the formula (5.22) is made inordinately complicated because of the way in which Martzloff has interpreted the change of signs between the two halves of the columns Δ and C. The exposition is not made any clearer when he chooses to continue the account by relying on a procedure devised by Uchida, instead of focusing his explanation on the steps in the Chinese text.
There are two points here. First of all it would help if the Chinese text were quoted, in parallel with the corresponding algebraic steps, just as Qu Anjing showed in his account of interpolation in Chinese Calendar and Mathematics (中 国 历 法 与 数 学, Beijing 2005, p. 240).
Second, the scientific context needs to be explained. The problem here is that the equation of the Sun is assumed to be zero at the solstices, whereas in fact it is zero at the apogee and perigee. Unless I am greatly mistaken the Chinese did not recognise the role of the solar apogee. The resulting solar equation will inevitably be in error, and no amount of fussing over the weakness in linear interpolation will overcome this. The situation in Chinese astronomy might be contrasted with the earliest development in Greek astronomy when the apogee of the Sun was determined by Hipparchus by an argument that must be seen as a stroke of genius.
In applying a similar equation to the Moon the Chinese procedure avoids making this error because the Chinese recognised the anomalistic month (in effect, the lunar apogee) and calculated the interval between the given lunation and the beginning of the current anomalistic month (ruli 入 曆). From the point of view of clarifying the scientific context, however, it would still help if one could know why the two corrections (solar and lunar) can be added in determining the final correction to the time of lunation.
Although the subtitle of this volume refers to “temps mathématique caché, linéaire et continu” this underlying scientific principle is not made explicit. We can make it explicit by observing that the mean positions of Sun, Moon, Lunar Apogee, Lunar node, and the five planets can be calculated by finding the interval in days from the Shangyuan and multiplying that interval by the daily motion of the Sun, etc. The initial values of these quantities is 270°, but that of course expresses the conversion to our equatorial coordinate frame. In Chinese terms we have a coincidence of all 9 mean longitudes at the Winter Solstice, illustrating the absolute centrality of the Winter Solstice in their theory. The Julian Date of the Shangyuan of the Xuanming li was noted above, ‑2580308748.5. Each new calendar will have its own Shangyuan, very different from the Indian practice where, in spite of changes in the parameters over the centuries, time is always measured from the Kaliyuga. In another very important calendar, the Daming li (of the Jin Shi), it is ‑32372914549.5.
The comparison of the Chinese mean longitudes against the modern is efficiently performed by plotting the deviations provided by the program Devplo which is available on my website www.raymondm.co.uk. The ‘deviation’ is the ancient mean longitude minus its modern calculation. The plot displayed here shows the deviations of the Daming li (of the Jin Shi). It is clear from this set of deviation curves that this calendar is based on a particularly coherent set of parameters. The date of this calendar is 1180 Jan 29, and evidently the curves, especially those of the Sun and Moon, show that there was good agreement with the true situation in the 11th and 12th centuries, leading up to the date of the calendar.
When calendars are analysed in this way we find that usually the Sun and Moon at least have been fitted closely to the date of composition, but that the other parameters might poorly reflect the true state of the sky around the time stated in the canon. It is clear that over the centuries the never ending revisions of the calendars were motivated by the careful insistence on agreement with observations of the Sun and Moon. With such an accumulation of records over a long period people were provided with an opportunity to average out over this long time base, and so to achieve a better value for the tropical year. In fact the length of the year is always slightly too large, so it would appear that the Chinese astronomers did not make really effective use of the long time-base covered by their records. However Guo Shoujing, author the Shoushi Li Calendar ca 1280, did undertake such a comparative study, and this led him to propose a secular change in the length of the year.
Deviation curves for the Daming li (Jin Shi j.21). The symbols have their usual astronomical meaning, and Γ represents the lunar apogee.

These critical observations are not meant to detract in any way from the volume under review, and we must be grateful to Martzloff for an account which presents in a western language for the first time a real introduction to the astronomical procedures underlying the Chinese calendar.
