The Azes Era

A new discovery and a new problem

Kushan chronology has recently been upset by the discovery of an inscription dated in two eras. Though the full inscription is still unpublished (Salomon, forthcoming), the text of its translation is available (Senior, 2003). The 'Greek era inscription' (insc. 23), written in Kharoshti on a reliquary reads as follows:

In the twenty-seventh - 27 - year in the reign of Lord Vijayamitra, the King of the Apaca; in the seventy-third - 73 - year which is called 'of Azes', in the two hundred and first - 201 - year of the Yonas (Greeks), on the eighth day of the month of Sravana; on this day was established [this] stupa by Rukhana, the wife of the King of Apraca, [and] by Vijayamitra, the king of Apraca, [and] by Indravarma (Indravasu?), the commander (stratega), [together] with their wives and sons.

This is a hugely important inscription. It means that for the first time we have clear evidence that a 'Greek' era was used in Kharoshti inscriptions. It also gives a way of calculating the era, take the Azes era (58 BC by convention) and subtract 128 years (186 BC give or take a year). However, a number of Kushan specialists have tried to apply this new date and feel it does not fit well with the date of 127 AD proposed for the era of Kanishka I. They would like to adjust by about 10-15 years. One answer would be to abandon the date of 127 AD, which is compatible with the present understanding of the date of Kanishka. However, for scholars who take Kanishka's date to be fixed at 127 AD, a more controversial approach has been suggested: move the Azes era.

The Azes Era

"Azes I ... was the founder of an era, and this era can be associated with reasonable certainty with one that later came to be known as Vikrama, on chronological and archaeological grounds" (Salamon, 1998: 182)

The statement by Salamon is probably a fair reflection of the confidence that most scholars before the recent publication placed on the equation Azes = Vikrama = 58 BC. At a recent symposium at Oxford the suggestion of moving the Azes era was proposed. Curiously, while everyone 'knew' the Azes era was the Vikrama era, no-one could actually recall evidence for this equation. It is clear that the date has to be close to 58 BC (a number of inscriptions in the era can be roughly dated) and everyone could recall that there had been articles (notably by Bivar & Fussman) which settled the matter in the late seventies, and early eighties.

Puri (1977: 27-9) attributes the theory that the Azes era was one and the same as the Vikrama era to John Marshall. Marshall (1914: 973) attributes the idea to Fleet. An examination of Fleet's articles in the Journal of the Royal Asiatic Society (1907: 1023, 1913: 1000, 1914: 797) shows that he actually proposed the basic principle that inscriptions should be fixed to known eras where-ever possible. Marshall took this principle and the Gondophares (insc. 353) and Panjtar (insc. 359) records which he knew had to be contemporary with the early Kushans, subtracted the dates (103 and 122) from the middle of the first century AD and got an answer roughly in the middle of the first century BC. He then attributed this to the Vikrama era which was close. Further, he published the Silver Scroll from his excavations at Taxila (insc. 360) and correctly read 'ayasa' as Azes. From this he argued that the Kharoshti documents were dated in an era founded by the Indo-Parthian Azes and that for the sake of argument this era should be considered equivalent to the Vikrama era. He concluded by saying (on the subject of the actual era date):

"... the identity of the era of Azes and the Vikrama era can hardly be regarded as fully established, and to my mind, it is quite possible that the era of Azes will be found to have commenced a few years earlier or later than 58BC" (997)

Curiously it was not the dating of the Azes era to 58 BC which attracted the greatest criticism. Instead, critics attacked Marshall's suggestion that the era was named Azes from Konow (1929:lxxxv), through Lohuizen (194 :chapter 1) and Basham (1953), and as recently as the 1960 London conference (Narain: 196) it had been strongly argued that the Azes era did not exist. These scholars suggested that the word 'ayasa' could not refer to the king Azes and must simply be an appellation of the month or other minor variation on the dating formula. The insinuation was that the Azes era did not exist. Freed from the constraint of linking the kharoshti inscriptions to Azes I these authors proposed various alternative dates (under the name 'Old Saka Era'), mostly earlier than 58 BC (Konow 84 BC, Narain 88 BC, Lohuizen 129 BC).

Consultation with Bivar's article 'The Azes Era and the Indravarma Casket' proved enlightening. Bivar never justifies the date of 58 BC. Rather he is concerned to demonstrate that an era associated with Azes actually existed. In the historiographical tradition elucidated above this makes perfectly good sense. The Indravarma casket did demonstrate clearly that tge Azes Era was real, and not a figment of Marshall's imagination. What it did not do was demonstrate that the link Azes = Vikrama = 58 BC was sound. In essence Marshall's warning quoted above still held good. Despite this, from the eighties onwards the date of the Azes era was taken as a given. Exactly why is unclear, but perhaps simply because the demonstration that Marshall had been right about the name of era lent weight to his opinion on the date.

It is now an opportune moment to examine the problem which this debate had been intended to solve...

Kharoshti Inscriptions
Insc. Date   Insc. Date
341 58   168 134
342 60   360 136
345 68   560 157
344 68   559 157
23 73   361 168
346 78   362 187
347 81   364 191
351 100   409 200
352 102      
353 103      
356 111   569 303
357 113   410 318
358 117   415 359
359 122   423 384
558 127   432 399


The Kharoshti Inscription Sequence

The table on the left shows a selection of dated Kharoshti Inscriptions excluding those of the Kushans. The inscriptions, are ordered simply by their date, and can be seen to provide a coherent sequence. Not as complete as the Kushan sequence (some gaps are as large as 19 years) but still very clearly a sequence. What will be noticed is there is a gap of at least one hundred years (some authors make it larger as they do not accept the dating of all the inscriptions from 150 to 200). Also, a small number of these inscriptions are clearly marked (in green here) as being in the Azes era. Two odd inscriptions are marked (one clearly stating it is in the era of Moga, and the dual dated inscription mentioned above, given in the table as its Azes date).

Two solutions to this sequence have previously been proposed. One is to date the entire sequence according to one era (57 BC, 83 BC, 88 BC, 129 BC). The substantial gap is then filled with the Kushan era, which runs to between 100 and 140 years in this region.

The other solution is to date the inscriptions in two sequences. One, the Azes era the other the Greek (150 BC,155 BC, 158 BC, 170 BC, 186 BC, are various suggestions). This solution assumes that the high dated inscriptions actually represent a shift to a new era (the same era as Vima Taktu's central Asian inscriptions [insc. 500 & 508] of 279 and 299) and the gap is the dislocation produced by that shift.

Our new evidence shows clearly that the first solution is impossible. And while it shows some form of the second must be true it also shows that previous relationships between the era's have been incorrect (positing as they do a gap of 90 - 110 years, as apposed to the 128 we now know to be the case). The new evidence also introduces extra problems. If the Greek era supplanted the Azes during the conquests of Vima Taktu (Greek era c.280, end of the first century AD) as is commonly presumed, then why is the dual dated inscription so early, in fact at the start of the sequence? And if the Greek era was used alongside the Azes for nearly a century, then the gap which to be explained by the shift from one era to the other should not exist?

The problems that need to be answered about the Kharoshti sequence are therefore:

  • In what year did the era of Azes commence (it follows from this that the year of the Greek era will then be known)?
  • Which inscriptions are dated in the Azes era, and which in the Greek era?
  • What is the era of the Moga inscription, and is it unique, or are other inscriptions to be dated in this era?
  • By what mechanism is the substantial gap (year 200 to 303) to be explained?
  • Are the high dated inscriptions, year 303 onwards, to be dated in the Azes era, the Greek era, or some other era?
  • How do the dates of the Kharoshti inscriptions relate to the anomalous high dates from Mathura (see below)?
  • How do the dates relate to the Central Asian dates of Vima Taktu?


The Vikrama Era

The Vikrama era begins in 58 BC. We know this because a very late text, Meratunga's Theravali gives a story of how King Vikramaditya expelled the Sakas from Ujjain and established the era, and gives a formula for calculating the Vikrama era from the Saka (Majumdar, 1951). The term Vikrama is fairly late, and was not the original appellation of the era. Before the ninth century the Vikrama era was known as the Malwa era. This Malwa era can be traced to a group of inscriptions of the fifth century, found to the south of Mathura in Rajasthan. These inscriptions include the word 'Krita' and this allows them to be linked to a group of inscriptions dated 295, 284, 282 (ie the first half of the third century) from the same region. This is where the trail stops. These are the oldest inscriptions known with certainty to be dated in the era of 58 BC.

The problem with the Azes inscriptions being dated in the same era is that there is no continuity. The Azes era seems to stop in use in the north-west by the year 200 (possibly earlier), and then the Krita era begins in Rajasthan, a considerable distance away nearly a hundred years later. There is therefore, no tangible link between the Kharoshti dates, and the sequence which came to be known as Vikrama.

Interestingly there are several anomalous high-dated inscriptions at Mathura (insc. 333, 32 & 334) dated in the years 270, 280, 299. These have in the past been attributed either to the Kushan period, and are considered to be dated in the Greek Era and precede Kanishka, or to be dated in the Gupta era and follow the Kushan period. The first explanation looks plausible for 334 which mentions royal titles, but implausible for 32 which uses the epithet 'sakya' for a Buddhist Nun. On the other hand the Gupta dating has never been very convincing, even for Luders who first proposed it. Of course, it is possible that different explanations apply to the inscriptions and their similarity in date is simply a co-incidence.

An alternative explanation is that these three dates are Krita dates, transferred from Rajasthan to Mathura in the period 217 AD to 246 AD (in the reigns of Vasudeva and Kanishka II). And thence from Mathura to Gandhara where our first date 303 corresponds to 249 AD (and the reign Vasishka). This would then explain the late inscriptions of Gandhara as being in the Krita (Vikrama) era and would provide a possible explanation for the large gap (the displacement of the Azes era by the Kushan, and its further displacement by the Krita).

Unfortunately, this explanation, like the Gupta or Greek assignment of these three inscriptions is unsatisfactory. It is raised here to show that there are a considerable number of possibilities in any explanation, and that there is more anomalous data to explain than some surveys often make clear. 

The Greek (Yona) Era

One method of solving the problem is to establish the start date of the Greek era. This can then be used to establish the Azes era using the relationship above (Great Year = Azes + 128). Unlike the Azes-Vikrama equation there is no known era which provides a good candidate for the Greek. So it is necessary to establish the Greek era's start date by some other method.

One that has been suggested is to establish which Greek ruler instituted the era and thus date its inception. A new source may be discovered which gives information about the inception of the Greek era but at the moment no such source is known. Without such a source there are two serious objections to this method. Firstly, the Bactrian kings cannot be accurately dated so even if one can be identified as the founder this would not date the era any more accurately than it is already. Secondly, the motivation in founding an era and the reasons that private individuals choose which era to use are not understood. So there is no reason for preferring any Greek King over any other for the start of the reckoning. Given these two objections it seems absurd to try and guess at the foundation of an era in the absence of any evidence. Even if the second objection could be overcome the first is fatal; both eras can already be linked to kings whose reigns are known as accurately as those of the Greek Kings of Bactria.

The only effective method of establishing the starting date of the Geek era is to find a date in that era that can be linked to a date in another era. Either, by finding a source (like insc. 23) which has dates in two eras (but unlike 23, in which one of the eras is known). Or by finding an inscription in the era which can be linked to some person or event which is already firmly dated.

There are very few such inscriptions. The only Kharoshti inscription which is certainly dated in the Greek era is the dual dated inscription 23. No method of distinguishing inscriptions in the Greek era has yet been established. What are required are inscriptions knwon to be dated in the Greek era, that can be fixed by some method other than the date. One group are the Central Asian inscriptions of Vima Taktu, written in Bactrian and dated 279 and 299 [insc. 500 & 508]. Unfortunately Vima Taktu's dates are known only roughly. He was succeeded by Vima Kadphises and then by Kanishka. Kanishka is the fixed point in this discussion, though his date is not known accurately. However it can be said that it is now agreed Kanishka falls in a fairly narrow range of dates c.120 AD. An additional point of uncertainty is the length of Vima Kadphises reign (though evidence still unpublished indicates it might be quite short). So while no accurate date is possible it is reasonable to place Vima Taktu in the late second half of the first century AD (and possibly into the second century). If we subtract from c.90 AD (a guess) the value 279 we arrive at c.190 BC. From which we get an Azes era of roughly c.62 BC (give or take the uncertainty already mentioned).

The Azes Inscriptions

King Azes Era Date
Gondophares, Kajula?  103
Un-named 122
Un-named 136
Vima Taktu 151 (279)
Vima Taktu 171 (299)
Vima Kadphises (insc. 362) 184/7
We will now proceed to do the same things with the Azes inscriptions. Three of these are relevant; Insc.353 (azes 103) 359 (azes 122) & 360 (azes 136). The first is dated in the reign of Gondaphares, and may mention Kajula. The second two are in the reigns of unnamed Kushan kings. They belong to the reign of Kajula, or (less likely) to the early reign of Vima Taktu (because they proceed the inscription dated year 279). On the left all these dates are converted into the same era (the Azes).

 Using this sequence it is possible to calculate the Azes era  (using any of the four kings mentioned) in much the same way that was done with the Greek era above. It points to the same basic conclusion that the date is in the middle of the first century BC. However, the margin of error is still quite considerable.


I have not supplied an answer to the problem of the Azes date. That, of course, wasn't the intention. The purpose was to survey a problem that is central to problems of Kushan Chronology, and that will not yield to simple answers. It is possible that a much more detailed study might yield a solution, or that fresh evidence might resolve some of the problems. A combination of both has reduced the margin of error on the Kanishka era to less than the margin of error on the Azes. This is why the Azes rather than the Kanishka era is likely to be the focus of chronological debates in Kushan studies in the next few decades, the uncertainty is now greater and thus its solution more profitable. I have already outlined what the problems are that need solving. Perhaps it will be useful in conclusion to outline what is known:


There is a simple rule of thumb in epigraphy (and source criticism in general) that if everyone agrees on what an inscription says then nobody really cares. Of the six datings given here only one is undisputed; that of Vima Taktu to 279, because it doesn't conflict with anyone's theory. Of the others it is disputed if insc. 353 names Gondophares or his successor Gondophares-Sases, and if 'erjhuna kapa' is, or is not, Kajula. In the case of both the unnamed Kushan kings (insc. 359 & 360) it is a source of controversey which Kushan king is referred to. The reading of 299 on insc. 508 by Harmatta is seriously doubted. And a debate on whether insc. 362 should be dated 184/7 in the Azes era or 284/7 in the Greek era has raged for nearly a century. It is therefore wise to be cautious about readings based on single inscriptions.

Contents Page The Date of Kanishka List of Inscriptions Huvishka Bibliography

Robert Bracey.