skip to primary navigationskip to content

Stage Three Project Databases

Description of the Stage Three Databases

The Stage Three Project's two databases, one faunal and the other archaeological, were conceived to be complementary, and indeed were constructed using many of the same sources. Only European archaeological and faunal assemblages which carry absolute determinations (radiocarbon, thermoluminescence, uranium-series, electron-spin resonance, etc.) have been included in our databases. Sites which have only been dated by relative methods are excluded, as we could not be sure that they had been attributed to the right period. The radiocarbon dates have been calibrated using the CalPal program devised by Dr. Bernhard Weninger and Dr. Olaf Jöris (as it is the only program known to us which can deal with determinations of up to 50 ka), enabling them to be compared to other absolute techniques which generate age determinations in calendrical years. For reasons of space, we have just published one set of calibrated ages, using the recommended CalPal curve (U/Th, Foram-GISP2-synchronisation and Lake Suigetsu varves), but have also published the original uncalibrated dates, so that users can calibrate according to their own requirements. 

Information has been gathered from the printed literature, with dates often being gleaned from site-reports, archaeological and faunal syntheses, and from datelists in journals such as Radiocarbon and Archaeometry. The databases are not by any means exhaustive: they were constructed to answer specific questions which could be linked with the modelling output and validation data being produced in the Stage Three Project. Hence, variables have been selected to test the presence/absence of humans and other animal species in Europe. The patterning obtained by mapping these distributions across time and space can then be compared with the model output and validation data generated by other members of the Project. 

Users planning publication should acknowledge the source and inform the contact addresses listed with the databases and climate simulations. 

Validation and Verification of the Databases

We have verified and checked the information in the databases as far as possible within the time-constraints which were available, but would be very grateful if users could let us know of any mistakes contained therein as soon as possible. The contact e-mail addresses are given under the relevant database sections below.

It should be remembered that many faunal sites have only been dated because they are also archaeological ones; there appears to be relatively few dated faunal sites with no associated archaeology. Archaeological absolute determinations which are derived from identified animal bones and teeth (including those on hominid fossils) are important in the faunal database, as they can be treated as direct, rather than indirect, dates.

All efforts have been made to position sites accurately. Detailed maps of site positions were used whenever available to locate the latitude and longitude (given in decimal form), but sometimes the available information was not sufficient to permit a high precision. Where this happened, the grid-references are given in square brackets. All site positions were plotted in MapInfo using the Bartholmew's digital map of Europe, in order to obtain a consistent set of grid references. While compiling the databases, we frequently discovered discrepancies in the positions given in the published datelists, often giving varaints several degrees apart; one site was moved from one hemisphere to the other between successive datelists by the same dating laboratory, and hence it became imperative to have a reliable set of positions.

Map Plotting Routine for the Archaeological and Faunal Databases

The Archaeology and Mammalian Fauna databases are equipped with a facility for the simple plotting of their contents on a geographic grid with an outline of Europe. The same grid is compatible with the output grid of the Penn. State climate and vegetation simulations, and can, for instance, be used to plot archaeological or faunal data on transparent overlays to match climate and vegetation maps.

The mapping utility has been constructed by and Lawrence Rush, of the Department of Earth Sciences, Cambridge University, on a shareware basis. The Stage Three Project is most grateful for their generous assistance, which makes the use of geographic aspects of the databases much simpler and more convenient. For Project members authorised to use the Limited Access section of the website, fees have been paid by the Project for multiple use.

Users can download a copy of the Instructions for Use of the map plotting program. They are in rich text format (Mapping Instructions), and can be printed out for subsequent reference.

The zipped folder contains atlasnow.exe, avshape.exe, atlas.GID, atlas.hlp, atlaswin.log, batch.bat, bwcc32.dll, stage3.prm, Wvssrcd4.atl and Wvssrcd44.Sc0. It is recommended that when users download these ten files, they save them on their hard-drive in a folder called something like "Atlas", e.g. C:\Atlas\..

Note: When users click on the hypertext link above, they should agree to the use of the evaluation-only use of the WinZip program. They can then use the zip wizard to unzip the folder and its contents.

The Archaeological Database

Guidelines by William Davies (August 2000)

The text (Guidelines) of this section can be downloaded as an MS-Word file, and it is strongly recommended that users do so, as extra bibliographic material is supplied in its appendices. I would be grateful if users could inform me (Dr William Davies) both of any mistakes they find and of any suggestions for improvement of these guidelines.

General Notes

The database contains almost 1900 determinations from about 400 sites, so there is much scope for error. It has been constructed with specific criteria and intentions in mind, which were designed to fit in with the output of the Stage Three Project. The two most important attributes for this purpose were the temporal and spatial distributions of human activity and skeletal remains in Europe.

I have included determinations in the database which are clearly not OIS-3 in age, and this is to enable users to assess the general levels of dating accuracy and precision for any given context. As absolute dates are based in part on statistical methods, it would be surprising if there were no outliers. Such dates may not necessarily be "wrong", but may instead reflect disturbance at the site: a piece of bone dated to 14,000 BP in a context otherwise attributable to 30,000 BP may really be that young and have arrived in that position post-depositionally. All known determinations from the assemblages listed in the database have been included, as I believe the analysis of the distributions of dates from each context to be essential to our understanding of the Palaeolithic record. It should be borne in mind that our spread of dates for any given context may not be complete, for it is conceivable that we may be missing some "outliers" owing to suppression.

The Notes/Bibliography column contains references specific to the age determination itself, together with any pertinent background literature on the nature of the site itself which could be added under the time-constraints involved in the construction of the database. The most frequently-referenced laboratory datelists have been reduced to an indication of the number of the datelist in the database, together with the relevant page numbers. The Groningen (abbreviated to "GrN"), Lyon, Gif and other datelists were published in the journal Radiocarbon, while all those associated with the Oxford Research Laboratory for Archaeology and the History of Art (abbreviated to "OxA") were published in Archaeometry. Full references for each of these abbreviated datelists can be found in Appendix 2 to the rich text version of these Guidelines, and it is recommended that users print out all the text in this document for ease of use.

In accordance with convention, editorial additions/interpolations and emendations are given within brackets, i.e. "[…]". Parentheses, i.e. "(…)", are not indicative of any editorial action taken.

Site Positions

The GIS program MapInfo 5.0 was used to plot the positions of sites on a Bartholomew's map of Europe. The latitudes and longitudes were measured with no map projection being imposed. MapInfo can convert these data into specified projections, e.g. Lambert conformal projections, if required. This "neutrality" in the co-ordinates will allow users of the database to convert them to their own specifications, and permit their use with other maps and map projections. Early on in the Project, it was decided that all latitudes and longitudes should be consistent, especially as it was quickly discovered that co-ordinates attributed to sites in the published literature (e.g. in Radiocarbon) were often wildly inaccurate, and sometimes the same site was even accorded different positions by different dating laboratories! Digital degrees have been used, with longitude preceding latitude (cf. the traditional order), and users should remember that longitudes in the western hemisphere are prefixed by a minus-sign, while those in the eastern one are rendered as positive.

Co-ordinates are given at three levels of precision. The most precise positioning is seen in the degrees given to four decimal places, while those given at three places reflect a lower level of confidence. Triangulations given in brackets have an element of uncertainty attached to them, either owing to the level of description in the sources or to disagreement within the latter. Their position is not hugely inaccurate (not by more than a few kilometres), and reflects the precision of the information given in the sources. In many cases, the co-ordinates in brackets refer to the position of the [nearest] settlement, with the number of decimal places reflecting levels of confidence. Bracketed co-ordinates given to four decimal places can probably be taken to be accurate, although some disagreement in the sources renders the brackets necessary. Users are advised to remove the brackets if they wish to plot the positions of these sites, but should always bear in mind that their precision is not [usually] comparable to those given without brackets. All possible attempts have been made, within the timescales available, to reduce the numbers of bracketed co-ordinates, but a residuum remains, especially the former USSR, where the available maps are on such a scale that precise plotting becomes well-nigh impossible.

Site Characteristics and "Cultural" Attributions

It was felt important to note the site-type for each site collected, so that any biases arising from a preponderance of sheltered (cave or abri [=rockshelter]) sites in relation to open-air ones could be addressed. If we are not aware of how many dated covered sites we have and use, we are inadvertently mapping the distribution of limestone in Europe. If the total number of known Middle and Upper Palaeolithic sites were to be considered, it would be apparent that open-air sites are actually remarkably well-represented. We have relatively few dates from these contexts because their preservation conditions are normally poor, especially in gravelly or sandy deposits.

The layers and "cultural" ascriptions for each date are given as they appear in the literature. Discrepancies, rather than obvious transcription errors, are noted in the "Notes/Bibliography" column, although it often becomes difficult to accommodate every researcher's cultural-taxonomic preferences, especially in areas where "splitting" rather than "lumping" research traditions predominate. This problem is especially apparent in the study of the early Upper Palaeolithic from eastern Europe and the former Soviet Union. Most taxonomic disagreements rest with the niceties of whether something is "Aurignacian" or "Gravettian", based on the proportions of certain types of tool. These criteria are often specific to individual researchers or research traditions, and I have tried to side-step this issue by preferring the majority view and then noting any alternative variants in the "Notes/Bibliography" column. Those users who wish to change and re-classify my attempts at classification are of course free to do so. However, it should be remembered that there is usually little dispute about what is attributed to the Middle and Upper Palaeolithic groupings (but cf. Samuilitsa in Bulgaria, which is variously attributed to either the Mousterian or the Aurignacian). The attribution of dates to layers can be extremely variable when one attempts to correlate accounts in the literature: in cases with several sources, I have tried to adopt the majority opinion, and have noted any variants in brackets in the "Layer" column and also under "Notes/Bibliography".

The term "FAUNA" in the "Industry" column refers to contexts with no certain archaeological content on sites known to have been occupied by humans. Some, like Paviland, were excavated in such a way that their published stratigraphy seems meaningless, and samples which show no evidence of human modification or do not derive from anthropogenic features, must be treated with suspicion. I have therefore elected to mark such samples from known archaeological sites as "FAUNA" if they are bones. Users of the database can incorporate them into the main body of more certain archaeological age determinations at their own discretion.

Absolute Age Determinations and Laboratory Numbers

As already stated, the absolute determinations in this database have been derived from the literature, and are thus subject to transcription errors made in our sources. Versions given in primary sources, such as laboratory datelists and site reports, have been favoured, unless corrections have been announced later. One difficulty encountered was the occasional divergence given between 14C determinations given in Radiocarbon, etc., and the versions posted on the BaNaDoRa Database (University of Lyon). As the latter source is constantly updated, it is sometimes difficult to know whether any variants represent corrections or transcription errors. Wherever possible, I have tried to favour the "majority" version if a determination exists in two versions in several sources (all variants are given in the "Notes/Bibliography" section, so that the user can make an informed choice). Errors are evident not only in the dates themselves, but also in the laboratory sample numbers, where transposition of numbers is all too easy (speaking from personal experience).

Non-radiocarbon determinations frequently do not have laboratory sample numbers, and often represent averaged results: further details will be found in the "Notes/Bibliography" column. Many 14C determinations have lower-case single letter suffixes, e.g. LE-1428a, LE-1428b, LE-1428c and LE-1428d, meaning that they are multiple estimates performed on the same sample. Some laboratories appear to have a greater predilection for this practice than others. Techniques such as ESR and TL often publish multiple age determinations on the same sample, and all efforts have been made to include all of these in the database (except when the [averaged] alternative provided proved to be sufficient). At a purely methodological level, giving all the variant determinations allows us to assess the quality of the material from the site and the taphonomic conditions to which it was subjected.

Dating Methods and Materials Dated

Various abbreviations have been used in the "Dating method" column:

  • "C14" = conventional (beta-counting) radiocarbon dating
  • "AMS-C14" = radiocarbon dating by accelerator mass spectrometry
  • "TL" = thermoluminescence
  • "OSL" = optically-stimulated luminescence
  • "ESR" = electron spin resonance: "ESR: EU" = early-uptake model; "ESR: LU" = linear-uptake model
  • "U-series" = uranium-series methods, including variants, such as "U/Th", etc.

When known, the materials dated have been given in a separate column, in order to assess the effect which substrate has upon the quality (accuracy and precision) of the age determination. Thus, care has been taken to distinguish "bone" from "bone fragments" and "burnt bone", for example. The species of plant or animal to which the remains once belonged are also noted, if known.

Correction of 14C Dates

Infinite dates could not be corrected, as they are essentially minimal ages, and have thus been left untouched in the "Corrected date" columns. The calibration program CALPAL has been used, as it can convert 14C determinations of up to 50,000 bp. The default calibration curve (U/Th, Foram-GISP2-synchronisation and Lake Suigetsu varves) has been used for this database in order to give users an idea of what the 14C determinations might look like in calendrical years. Database users who would like to obtain a copy of this program for themselves (perhaps in order to try more than one calibration curve) should contact either or , at the University of Cologne, for details. Non-14C determinations (TL, ESR, OSL, etc.) do not require correction to calendrical (solar) years, and have been left untouched.

Hominid Remains

The presence of hominid remains warranted the creation of a separate column, so that the skeletal data could be compared independently with the commoner archaeological material, and assumptions about bio-cultural associations questioned and assessed. Much material has been gleaned from Oakley et al. 1971 (Catalogue of Fossil Hominids, Part II: Europe; London, British Museum (Natural History)), but new material and interpretations have also been incorporated where known. I should be especially grateful if users could inform me of any omissions or mistakes.

The Faunal Database

There are many similarities with the archaeological database, as both were constructed simultaneously. Users of the faunal database are therefore advised not only to download a copy of the Table of Faunal Codes, but also the Archaeology database Guidelines before they start. Both downloads are in MS-Word format.


In the event of any problems downloading, please contact Dr William Davies.