Digital projects – like all scholarly products – are not perfect. Since their results are shaped by many biases, critical users of their results must be aware of these biases and take them into consideration. The following list is a partial list of such biases, many of which affect the numismatic scholarship upon which FLAME is founded.

  • Use bias – in late antiquity as today, certain coins are used more or less commonly than others. This phenomenon may be a result of different denominations, different purity, or even their aesthetic value. The differences in coin use would bias the surviving evidence, as coins used more commonly would be more prone to loss (and therefore would be found more frequently in excavations), while some coins that were used less commonly may be over represented in hoards.
  • Loss bias – certain coins were lost more frequently than others. This is partially a result of other biases, such as use bias, but could also be the result of the different values of coins. According to a frequent argument, contemporaries would spend more energy and attention trying to find a high-value lost coin (e.g. a precious gold coin) compared to a low-value one (such as a low bronze denomination).
  • Survival bias – certain coins survive better in the soil and preserve their features for longer. Certain soils are more acidic and cause coins to dissolve more quickly than others. Both effects create biases for the final patterns of coin distribution depicted in reference tools such as FLAME’s app. In particular, it is likely that in regimes of tri-metallic circulation, silver coins are likely to be underrepresented in survival by comparison to gold coins, frequently hoarded, or copper-based coins, commonly found in excavations.
  • Reporting bias – many countries in western Eurasia have laws in place that require finders of antiquities to report them and hand them over to the authorities. Some coins are more likely to be reported than others, although this depends to a great extent on the finder. Some finders may prefer not to turn over more valuable gold coins because of the financial incentive involved. Others may prefer not to turn over low-value bronze coins because of their perceived unimportance.
  • Excavation bias – excavations are more likely to be conducted in certain areas, and these locations frequently do not correspond to the economic importance of the area in late antiquity. Modern politics, permits, funding and ideologies determines which areas would be excavated, and within those, which chronological layers would be prioritized. These decisions determine the resulting circulation patterns.
  • Hoard bias – hoards would likely be more common in times and places that suffered some sort of instability, which would encourage the people who buried them to do so. Moreover, it should be emphasized that the hoards that are found are a subset of all hoards buried, representing those hoards that were not recovered by their owners for some reason.
  • Reference work bias – reference works are key resources for projects such as FLAME since they concentrate a large amount of scholarship in a relatively standard format, establishing agreed terminology (e.g. for denominations) and comprehensively surveying other types of data (e.g. mints). These reference works, however, are not evenly distributed spatially or chronologically. Some times and places (e.g. the Roman Empire) are far better covered than others (e.g. the Aksumite kingdom, or the Sasanian Empire). These create biases that favor the former over the latter.
  • Selection bias – modern academia includes many pressures and demands on scholars’ time, and these force scholars to sometimes make difficult decisions about where to focus their attention, and how to invest their limited resources. In practice, this means that certain layers or areas in archaeological sites would not be excavated, or that certain coin finds – which are too complex or not attractive enough – might not be published. Scholarly selection therefore consists of another bias.
  • Identification bias – Coins with distinguishing features such as the 40-nummi piece with its high-relief M are easier to identify and catalog than coins without such features. Mint marks are far easier to read on certain coins than on others, creating a bias in which the difficulty (or impossibility) to determine particular features on coins would make certain variables such as their mints disappear in published scholarship. In particular, coins of the fifth century, ‘minimi,’ are much more likely to appear as ‘unidentified’ in excavation reports, or not be listed at all, than coins of other eras.
  • Decentralization bias – numismatic publications are notoriously decentralized and come out in many languages, many of which are not traditionally used for international research. Although there are a few publications that attempt to report on recent numismatic papers, they are rarely comprehensive. Therefore certain numismatic publications would be better known and better represented in FLAME’s app than lesser known or untraditional publications.