The Era of Kanishka
|Kanishka before the fire alter|
Kanishka is the most famous of the Kushan kings, he is preserved in Bhuddist tradition as the king responsible for calling the second great Bhuddist council. His series of coins is also magnificent, though not as extensive as those of his successor Huvishka. His warlike exploits and the strength of his kingdom are remarkable indeed, and it is felt by most historians that his reign marked the height of Kushan dominance in central Asia.
Unfortunately it has proved beyond the grasp of all those who have studied Kushan history to actually decide when Kanishka came to power, in which year did he become King of the Kushans. This particular problem has inspired a lot of the material on the Kushan empire and even two London conferences on the subject, the first in 1913 and the most recent in 1960. Though without a successful conclusion.
Dates have varied hugely, 57BC, 78AD, 115AD, 128AD, 134AD, 144AD, 230AD, and
others. The reason it matters, is that inscriptions in India and Central Asia
are date for a hundred years in the era that Kanishka founded. Fixing
that era would provide a chronological assistance to art and political
historians interested in the history of North India and Central Asia..
Two competing methods for resolving the problem of Kanishka's era have plagued Kushan studies. The first, and most popular, is to attempt to find a 'killer' argument. A new piece of evidence, or a radical interpretation of a previous one, which clinches once and for all the date of Kanishka's era. The problem with such a method is that so many convincing arguments have been put forward for so many different dates, that no faith can really be placed in them. Buddhist legends (Fleet, 1913) were used to prove 57BC. Sequences of inscriptions (Lohuizen, 1949), to demonstrate an equivalence with the Saka era of 78AD. Chinese sources to demonstrate 115 AD (Narain, 1990). Comparison with Roman coins for 3rd century (Gobl, 1999). Other eras for 134AD (Harmatta 1994). The most recent such attempt is that of Harry Falk, in which he claims to have demonstrated the first year of Kanishka's era was 127AD. The problem in Kushan studies is not that we lack of evidence for Kanishka's era, but that we have too many compelling pieces of evidence.
The second method is to take the all evidence and attempt to find a solution which is compatible with as much of our sources as possible. This method was employed by Cribb (1997) to arrive at a date of 107 to 120. It was previously used on this site to arrive at a date of 115AD. It was also the philosophy employed by Rosenfield (1967) to arrive at his date of 115AD.
Cribb has since abandoned his dating for Falk's date of 127 AD. This may seem illogical, as readers might assume that the second method is superior to the first. This is not actually true, the first method is the one that has been employed to date numerous other eras (Greek/Yavanna, Azes, Saka, Gupta, Bactrian Letters). Unlike the second method it presents a fixed, absolute single year, whereas the second gives a range of dates with varying degrees of confidence. The first method is also simpler, its assumptions can be checked more easily, and so greater confidence can be placed in the conclusions. With the second method, errors are hard to detect. However, the first method depends upon not having contrary evidence, and the reasoning behind the solution being quite certain. Every such solution so far proposed has fallen a long way short of the goal. And with each new solution proposing a conflicting date the level of certainty that must be demanded of future solutions increases.
So here, I will employ the second method, of attempting a synthesis of the available evidence. However, I have further revised the method. Previously the objective was to find a date which was compatible with all the evidence. This is in practice impossible. No date can be proposed which is not incompatible with some of the evidence. Instead I am now proposing to examine the weight of evidence. In total I have taken 10 sources of evidence, which can be considered independent of each other. Some evidence has been excluded from consideration, either because it is too unreliable or because the range of dates it gives are very broad. I have then presented this information graphically below. The first chart shows the range of dates indicated by each piece of evidence (dark blue - most plausible interpretation, light blue - dates that are acceptable). I have then presented the period from 90 AD to 170AD in the second chart showing a year by year indication of how many pieces of evidence support that date.
The evidence on which these charts is based is presented below. A cursory glance at Chart 1 strongly suggests that it might be possible to find a date in the second decade for which all the evidence agrees. It is a tantalizing, but frustrating, promise. What the chart does show is the reason that dates such as 78 AD and 230 AD have become increasingly minority positions. Our solution, it cannot be doubted lies in the first half of the second century. Chart 2, which concentrates on the period 90 AD to 170 AD shows this clearly. At either end the support for a dating falls away rapidly. The chart suggests attention should be focused between 105 AD and 146 AD, with particular attention on the two apparent peaks, 110 - 115 AD, 124 - 139 AD.
Our method has excluded a number of previous dates, 78 AD, 100 AD, 144 AD, 230 AD, but with the exception of 144 AD these were minority positions. What it has not done is separate the two positions, c.115 AD and 128 AD (or Falks 127 AD), which have enjoyed considerable support in the last three decades. Instead, we have discovered that these polarised positions are a feature not of historical debate, but rather a genuine reflection of the sources.
This leaves in place the problem of Kanishka, but it also leaves in place a method by which it might eventually be solved. New evidence, or refinements of old evidence can quickly and easily be accommodated in these charts. In 2004 the date of Kanishka's era is restricted to two narrow bands within a 29 year period. The day on which we finally put the oldest problem in Kushan Studies to rest might not be to far off.
Sometime in the reign of Kanishka II, or very shortly afterwards the Kushan kings lose political control of Bactria and Central Asia. We know this because their coins cease to circulate in that region, and the assumption has always been that the region was conquered by the Sasanians, who go on to establish a Kushano-Sasanian dynasty in the region.
Since this event takes place between 98 and 129 years after Kanishka it would seem a very powerful tool for dating him. However, we do not know when the Sasanian conquest took place. The earliest possible date would be in the reign of Ardashir I (the first Sasanian king, 226 - 239), whom the Arab writer Al-Tabari tells us made extensive conquests in the east and received tribute from the Kushans. The era of the Bactrian letters, 233AD (Sims-Williams, 1997, 2002) might be referred to Ardashir's conquest. Alternatively the era may have been inaugurated later and Aradashir's success greatly exagerated.
The same applies to Shapur I (240-273AD) who is widely seen as the most likely candidate to have conquered the Kushan empire, though again we cannot be certain he is responsible. However, if Niktin is correct that the Kushan-Sasanians are ruling by the 4th century (and it is implausible that they are not) then the region must have fallen under Sasanian control by 300AD
The range of possible dates is shown between a possible conquest in 233AD, and the end of the reign of Shapur. 272AD. With the most likely date considered to be the first decade of Shapur's rule, 250AD.
Another estimate would also be possible if the dates of Kushano-Sasanian rulers could be fixed with reasonable certainly. To show this I have included line A2, which is based upon two coin hoards discussed by Joe Cribb (1981). These coin hoards are from the Gandhara region (modern Pakistan) and as such represent a further incursion by the Sansanians and Kushano-Sasanians into the territory of the Kushans. They include eight coins of Hormizd I (issued by his Satrap Kabad) which have been overstruck with designs of Vasudeva II. Cribb concludes that the date of Kanishka must be 100AD - 128AD.
Let us examine the evidence. Cribb has good grounds for giving the following sequence of Kushano-Sasanian kings in Gandhara:
|Piruz I - Hormizd I with Kabad - Hormizd I with Meze - Hormizd II - Piruz II - Unknown King - Shapur II with Kabad - Shapur II|
Of these, firm dates can be assigned to some. Piruz I is probably ruling in AD 242. Hormizd II must proceed 302AD. So Hormizd I with Kabad coins were circulated sometime from AD243 to 301AD, Cribb suggests c. AD276 but any time from AD250 to 280 would seem reasonable. Now it therefore follows that Vasudeva must have overstruck the coins after this period. Since there are eight coins, and no overstrikes of Hormizd I with Meze, it is reasonable to suggest that Vasudeva's overstrike was made at this time and therefore the two kings are contemporaries. Ideally, of course, we would like to see Hormizd overstriking Vasudeva II as well, but historians rarely get to pick their evidence. If we could now fix Vasudeva II's date relative to Kanishka I we would be able to make considerable progress. Unfortunately, that is no mean task, and the best we can do is take an educated guess at second era 65 to 75, allowing for the wider possibility of a second era date between 45 and 95.
It is intriguing to note that this tends to indicate a lower date than that indicated by examining the invasion theory.
The Gupta kings of Eastern India founded an era that begins in 319AD. During the reigns of the early Gupta kings they conquered large parts of Northwest India, and so like the Sasanians they provide a limit on how late we can place Kanishka.
The Gupta king Samudragupta (whose dates are uncertain but must be between 320AD and 375AD) claims on an inscription at Allahabad to have a subordinate Kushan king, named as Shaka. His successor, Chandragupta II, actually has an inscription at the city of Mathura (dated 380AD), which we know was still under Kushan control as late as king Vasishka and possibly as late as Vasudeva II. Chandragupta also mimics the coinage of Vasudeva II (Mukherjee, 1978: Plate XXXI, 15, Plate XXXII, 2), which further implies that his take-over of Mathura follows very late Kushan rule there.
I should be noted that the Guptas do not seize Mathura from the Kushans. They may well seize the Kushans eastern domains from Shaka, but the Puranas, and coin finds strongly indicate that a group of local kings ruled at Mathura before the arrival of the Guptas (Sircar, 1956: 171), and after the later Kushan kings. Unfortunately, how long they ruled, and whether they took Mathura from Vasudeva II or from his successor Shaka is beyond the rather meagre evidence available.
The upshot of this is to imply that there must be at least 200 years between Kanishka and the commencement of Gupta rule at Mathura, certainly at least 150 (which would require it was lost in the reign of Vasishka). Since we can also link Shaka with Samudragupta, he must have begun his rule post-320AD (certainly post 290AD). Linking Shaka to Kanishka is very difficult but it seems implausible, based on the sequence of Kushan kings that he could be more than 200 years later.
The Kushans wrested Gandhara and India from the dynasty of Gondophares. Unfortunately, like the Kushan-Sasanians, the Indo-Parthian dynasty has been the subject of heated debates surrounding the order and dates of its kings. Presented in the table above are two date sequences based on the Indo-Parthians.
The first, C1, assumes is based upon the Azes era of 57/8 AD. This is used in a series of inscriptions of the Indo-Parthians, and the date of the era is widely accepted (MacDowell, 1991: 243; Senior, 2004: 5; Cribb, 1997: 223), though that should not be taken as the same thing as certain. When examining data from outside Kushan studies care must always be taken that our lack of knowledge can make theories more certain than they really are. However, if we assume that the three inscriptions of an un-name Kushana (probably Kajula Kadphises) are dated to the Azes era this gives us 45 to 78AD (broadly in agreement with the coin evidence that will follow). This much is quite interesting but we then have no Kushan inscriptions in this region until that of Wima Kadphises, dated either 184 or 187. If this is also dated in the Azes era then Kadphises is still in power 126 or 129AD. Unfortunately, while the Azes era is the most popular candidate for these inscriptions it is not the only one, some Kharoshti inscriptions probably belong to a Greek/Yavana era (Senior, 2004: 5; Narain, 1966: 237), and in 1960 Narian dissented from the Azes interpretation, taking the years mentioned to be in the Pahlava era (Narain, 1966: 236-7). Despite these reservations we will take it to imply a date in the 130s, give or take some margin for error depending on how close we are to the end of Kadphises reign, and the uncertainty relating to eras.
The second line, C2, is based on the numismatic evidence for the Indo-Parthian to Kushan transition. The key fixed date is Gondophares, who is dated by the Takht-i-Bahi inscription from 23 to 46AD. This date is also supported by other evidence, and is the majority view (Cribb, 1997: 224; Alram, 1999: 38) but there are dissenting voices who would be willing to place Gondophares nearly 50 years earlier (Senior, 2004: 5).
Gondophares is followed fairly clearly by Abdagases, Sarpadnaes, and Sases (and possibly others). Kajula Kadphises overstrikes on the coins of Gondophares, and his successor Vima Taktu follows Sases in Northern India (and is overstruck by Sases Indo-Parthian successors such as Pacores). There is also some evidence that Kajula's coinage briefly follows Sases (Bopearachchi, 1999: 136; Cribb, 1997: 225), which makes his rule contemporary with the period from Gondophares to Sases, the period 46AD to 78AD (Alram, 1999: 45). If we could be confident of the period of rule from Gondophares to Sases, and the length of the rule of the two Vimas, then Kanishka's era could be confidently dated. All we can say is that Vima Takpiso ruled for at least 20 years, and it would be unreasonable to suggest a combined rule shorter than 30 years, or longer than 60 (the reign of Huvishka-Vasudeva). If this is appended to a rough date of 80AD then the period 110-119 looks most promising, but unfortunately the uncertainties allow a considerable period either side.
There are several sources of Chinese evidence. The first is from the city of Khotan. This, according to Chinese sources, was the subject of Kushan political domination during the period 107AD to 127AD. In particular the Hou Han Shu states that they imposed a new ruler on the city in AD115 (or possibly a year earlier or later). If it could be established which Kushan king was responsible for this then it would help in fixing the date of Kanishka.
Narain (1990:164-165) links the Khotan prince to a later Buddhist chronicle that names Kanishka as the king who was involved, and Cribb fits the Kushan coins, all of Kanishka, into this period (Cribb, 1984, 140-1). If Cribb and Narain are correct then Kanishka must have come to power between 105 and 116AD. However, it is possible that Kanishka is not the king responsible, in which case some other date would have to be proposed.
What is certain, is that the report on the western regions which describes primarily the activity of Ban Chao up to the last decade of the first century, talks about only two Kushan kings. These are probably Kajula and Vima Takpiso. This excludes Kanishka having come to power before the report is compiled, in 96AD, and implies that Vima Takpiso must have come to power after 96AD, so unless his and Vima Kadphises rule exceeds 50 years, very unlikely, Kanishka must have come to power before 146AD.
This information, which is the most that can be gathered from the report on the Western Regions, is presented as A1 in the table above. It should be noted, Buddhist records are unreliable, coins can move around for trade for many reasons (the extensive finds of Kushan coins in Eastern India indicate that Cribb is wrong to suggest small denominations cannot move for trade), and there is no way to confirm for certain who the two kings mentioned in the Hou Han Shu are.
However, there is an additional piece of information in the Chinese reports. A report of an embassy from the Kushans (referred to as the Ta Yu-Chi) in the third century. This is presented in the table as D2. The San-Kuo Chih (compiled in the third century) reports "On the day Kuei-mao (26 January 230AD) the king of the Great Yueh-chih, Po-t'iao, sent an envoy with tribute. (Po-)t'iao was made "King of the Great Yueh-Chih Affectionate Towards the Wei".
The problem with this passage is identifying the king in question. Chinese renderings of Kushan names are usually badly mangled, and though this kings name has been linked to Vasudeva, it is impossible to do so with certainty. However, on the assumption that the king in question is either Vasudeva I or Vasudeva II, line D2 has been reconstructed accordingly.
Imperial Rome, from Augustus to Marcus Aurelius, is probably the most studied period in human history, and without doubt the best studied in the ancient world. Though not all dates are beyond doubt most are so well understood that if a firm link between some item of this empire and a member of the Kushan dynasty could be established it would put us much closer to resolving the problems of Kushan chronology.
There have been no shortage of attempts. Robert Gobl attempted on stylistic grounds to draw a link with the regimes, coming first to a conclusion that supported a date in the second century, and later to his controversial claim of the third century. For the simple reason that stylistic evidence is never reliable enough to permit fine dating I will not even attempt it here. Instead let us examine two finds of Roman coins from the Kushan region (Macdowell, 1968: 141-143).
The first is at Manikyala, inside a Tope, one Kajula, two Vima Kadphises, and seven Kanishka coins (all copper) were found. Along with four Kanishka gold quarter dinars, and seven Roman Republican Silver Dinars. We know that Roman coins like this were in circulation for a long period of time, well into the third century. We also know that they were exported to India after 64 AD. The reason being that in 64 AD the emperor Nero reforms the coinage, reducing the amount of gold and silver in the coin. This creates a discrepancy, inside the Roman Empire all coins are worth the same (enforced by the central authority) but outside the empire coins are simply pieces of gold and silver. So it became profitable to export pre-reform pieces.
The second find is at Jalalabad, where seventeen dinars were found, ten of Vima Kadphises, six of Kanishka, one of Huvishka. Also present were three Roman aurei, Domitian, Trajan, Hadrian (in the name of his wife Sabina). The last of these coins can be dated between AD 128 and AD 137.
The problem is one of establishing what these coins tell us, considering the vagaries of the circulation. Well in the first case they cannot have been buried prior to AD65, and whenever they were buried Kanishka is likely to have been ruling. In the second case they cannot have been buried prior to AD 129 and this seems likely to represent the early reign of Huvishka. Our most likely period is prior to AD 138 (if the coins had been exported later, we would expect a later coin present). This is broadly in line with the period that other finds indicate these coins were being exported (MacDowell, 1968: 139-40), but a late date cannot be excluded.
So far we have looked at the evidence of how the era of Kanishka fitted into evidence from the other empires of this period, Rome, China, Indo-Parthia, Guptas, Sasanians. The Western Ksatraps ruled India in the regions of Madhya Pradesh, Gujurat, and perhaps the southern Indus. They dated both their inscriptions and their coins according to an era beginning in 78AD and usually known as the Saka era.
Considering that the Kushan Empire and the Western Ksatraps shared a border hundreds of miles long for the major part of the rule of both dynasties it is remarkable that they do not provide more evidence for the dating of the Kushans. There are in fact only two sites which are helpful, Sui Vihar where the Sutlej River joins the Indus, and Eastern Malwa, the region that contains Vidisa and Sanchi. These are shown in the map to right, which shows the territory claimed by the Western Ksatrap king Rudraman I. Note that both Sanchi and Sui Vihar fall inside the region claimed by Rudraman.
Some authors have considered Rudraman's inscription as decisive, because they reasoned that the height of the Western Ksatraps domains could not be co-incident with the height of the Kushan Empire under Kanishka. However, this presupposes that we understand the relationship of the two empires, which we cannot until we understand the respective chronologies.
There are two contradictions. First, Rudraman claims in his inscription of 150AD to have control of the southern Indus valley and to have conquered peoples as far north as the Sutlej River. Yet there is an inscription of Kanishka dated to year 11 from Sui Vihar (no.178). Secondly, the later inscription of Sridharavarman in year 200 (278AD) is inscribed at Sanchi, but we have two inscriptions of Vasishka (nos. 58 & 62) of years 22 and 28 from Sanchi, and Rudraman claims Eastern Malwa within his domains.
It is possible that Rudraman is simply not telling the truth about the extent of his domains, or that border towns changed hands often and quickly, or that the Kanishka inscription is an inscription of Kanishka II. Due to the uncertainty in the Kushan sequence of kings the second conflict is of little use. But if we assume that Kanishka's rule could not have coincided with the Rudraman's conquest (on the assumption it must have taken place in Wima Kadphises or Huvishka's reign) that rules out the period from 127AD to 150AD. It is not however, the decisive dating tool previous scholars have suggested.
I have discussed the evidence for this extensively elsewhere. The source provides two possible interpretations, if we take it as reliable (and there are reasons not to). It is either the year of Kanishka is 227AD, or a post Vasudeva era is established in this year. If that is Kanishka II's era and we limit any inter-regnum to 10 years then Kanishka is to be dated between 117AD and 129AD. If it is a post-Kushan era, then it places no particular limitations on the date of Kanishka, other than requiring it to be before 160AD.
It is important to understand the sequence of the Kushan kings, since this is central to all of the arguments. Firstly there are inscriptions of various years from year 1 to year 98, of three Kushan kings, Kanishka, Huvishka, and Vasudeva. This is the basis of our chronology. There is also a sequence of Kushan coins about which there is board (but not universal agreement. This sequence runs: Kajula Kadphises, Soter Megas, Vima Kadphises, Kanishka, Huvishka, Vasudeva, Kanishka II, Vasishka, Kanishka III, Vasu, Shaka.
Thanks to the Rabatak inscription, we know that Soter Megas corresponds to Vima Takpiso (or Taktu), for whom we have two Central Asian inscriptions dated in the years 279 and 299, so his reign lasted at least 20 years. In addition, several of the other kings are mentioned in inscriptions. There are inscriptions of Vasishka from year 20 to 28, and an inscription of Kanishka of year 41, which presumably corresponds to Kanishka III. There is also an inscription of year 170 of Vasudeva, which may correspond to Vasu, and a Kushan king called Shaka mentioned in a Gupta inscription. There are several inscriptions of an unnamed Kushan king, dated 103, 122, 136, which may well belong to Kajula.
If we assume that the reason we do not find inscriptions of Kanishka II is because the years on them coincide with those of Kanishka I, so we are simply unable to separate them, then we can assume that Kanishka II through to Vasu, form a second series of dates, in addition to those of Kanishka I. This could either be a continuation of the Kanishka I's era, or a new era founded by Kanishka II. The second interpretation is actually more likely but there is no reason to assume the gap between the end of one era and the start of another is very long.
This gives us the following sequence:
|King||Length of Reign||Dates of Inscriptions|
|Kajula Kadphises||min.23 years||103-136|
|Vima Takpiso||min.20 years||279-299|
|Kanishka II||aprox.19 years|
|Vasishka||8 - 20 years||20-28|
|Vasudeva II||unknown||170 (disputed)|
This table is central to our understanding of the Kushan dynasty in chronological terms. The most important data that needs to be extracted from the table is the length of time between any king whose date we can otherwise fix and Kanishka.