The stylistically Paleolithic petroglyphs
of the Côa valley (Portugal) are of Paleolithic age
A refutation of their <<direct dating>> to recent
times
João Zilhão
5.2. Watchman's <<post-hoc accommodative argument>>
In a letter to EDP dated January 19 (that is, four months before setting
foot on the Côa valley), in which he offered his services as a dating
professional, Watchman said how <<greatly disturbed>> he was
to know of the deeds of <<so- called archaeological experts professing
to know the antiquity of the engravings without carrying out any scientific
dating tests>>. He also made it quite clear to EDP that <<I
therefore do not accept the general consensus, that seems to pervade Europe,
that engravings and paintings of horses, bulls, etc., are necessarily 20,000
years old>>. Also, according to Fischman (1995), as soon as he arrived
in the Côa valley and began to examine the engraved panels, <<several
bells went off in my [his] head saying these things are young>>. So,
it is fair to say that Watchman began his work convinced that the engravings
not only were not of Paleolithic age but, instead, were probably very recent.
When he obtained results indicating minimum ages in the order of 7000 years
he was therefore <<puzzled>> (Fischman 1995): his expectation
that the art was modern, as well as his theory that the skins on surrounding
rock were older than those covering the petroglyphs, were contradicted by
such results. One might think, therefore, that he would proceed to reexamine
both theory and expectations or, at least, that he would use a certain amount
of prudence and modesty in the presentation of his interpretation of the
chronology of the petroglyphs. What he did, instead, was to issue categorical
statements as to their very recent age, based on what is a text book example
of what archaeologist Lewis Binford (1983) has named <<post- hoc accommodative
argument>>.
The first step in this argument was that of dismissing the dates on the
skin covering the petroglyphs themselves as too old due to contamination.
His microscopical examination of these skins showed that they were made
up of a silty brown material <<probably eroded from the hillsides
above>> (Fischman 1995). This material contained particles of graphite
weathered out of the rock, and graphite, <<formed from ancient carbon>>,
would have been the contaminating material -- <<it made the engravings
look anomalously old>> (Fischman 1995). However, if the graphite was
indeed weathering out of the rock, it should also be found in the silica
skins covering the surface adjacent to the petroglyphs. But, according to
Watchman (1995a), that was not the case: the latter are described by him
as <<uncontaminated>>, as opposed to the silty crusts inside
the grooves defining the petroglyphs, which are described as <<contaminated
with ancient graphite and charcoal>>. No explanation is provided,
however, for the apparently contradictory situation that arises from this:
graphite was present in a crust formed by the accumulation of silty material
coming from the erosion of the surrounding soils; but it was absent from
the silica skins precipitated by flowing water over the surfaces of rocks
where graphite is a natural component!
This brings up the question of whether the graphite Watchman found in the
grooves may not have a totally different origin. One possibility immediately
comes to mind: that the graphite was introduced in the grooves in the context
of the several procedures, using different materials (pencils, chalk, paint
and wood charcoal), that were used by visitors to enhance the pictures before
the Côa art sites were fenced. Since graphite is a component of pencils,
its presence only in the grooves and not in the adjacent rock surfaces might
be interpreted as a clue to the contamination of Watchman's samples by young
carbon (particles of wood from the pencils, for instance), as well as old.
On the other hand, the fact that many such episodes of enhancement took
place at all the sites sampled in the framework of EDP's <<direct
dating project>> obviously questions the basic premise of Watchman's
approach -- that of the integrity of the 0.1 mm thick crusts that he analyzed
and sampled. Incidentally, that fact also provides the explanation for a
pattern observed by Watchman, which he mistakenly interprets as indicating
that the engravings are of a very recent age: the absence of lichen cover
in the grooves as opposed to its presence on adjacent rock surfaces. In
at least one instance, that of the panel 6 of Penascosa (Fig.
4), the present author can testify that such cover was also present
in the grooves when the panel was first discovered in late January 1995.
In any case, given what Watchman himself had previously stated on the implications
of such enhancement procedures, it is quite clear that, under the circumstances,
AMS radiocarbon dating should never have been attempted in the first place:
<<any form of chemical addition to the surface, especially paint,
will significantly affect the chances of ever obtaining a reliable radiocarbon
date for carbon-bearing substances that may be present in the engravings>>
(quoted from a letter dated January 19, 1995, sent by Watchman to the board
of directors of EDP). Bednarik, Watchman's co-participant in EDP's <<direct
dating project>>, seems to be, or have been, of the same opinion:
<<the introduction of foreign carbons, by any means and in any amounts,
renders AMS radiocarbon dating invalid>> (Bahn et al. 1995:31).
It should also be noted that Watchman's description of graphite as a <<contaminant>>
contradicts his characterization of the age of the samples. If the petroglyphs
are about one hundred years old, as he asserts, the crusts formed over them
are therefore younger. But if, as a result of contamination by graphite,
those crusts formed less than a century ago provide, as happened in at least
one instance, radiocarbon ages of up to 7000 years BP, then it follows,
given the laws of radioactive decay and the half-life of 14C, that Watchman
should describe his samples not as made up of modern organic material <<contaminated>>
by graphite, but as the opposite. The carbon contained in them would have
to be almost entirely made up of graphite <<contaminated>> by
very small amounts of modern organic material! In the case of the <<7000
year old>> sample, graphite would have to represent 98% of its total
carbon content!
Now if, as Watchman said in his presentation to the September 1995 International
Rock Art Congress in Turin, the <<organic matter>> in the mineral
accretions that he analyzed is essentially made up of diatoms encapsulated
by silica, how does he explain, then, that the dated samples turn out to
be essentially made up of graphite? In other words, Watchman is facing here
what seems to be an inescapable dilemma: either his sampling procedures
are adequate and the contamination cannot possibly be that extensive (and
then the silty crusts are much older than he thinks); or they are indeed
as young as he thinks they are (and then his sampling procedures are in
clear need of substantial improvement). That such improvement may indeed
be necessary is indicated by Watchman's statement on the specific locus
of the graphite contamination problem: <<graphite (...) occurs in
thin yellow-brown weathering rinds at the base of the silty brown accretions>>
(Watchman 1995b). Since he had told us before that the dates obtained referred
to the silty brown accretions themselves (<<carbon from silty brown
accretions developed in engravings gives ages ranging from 3000 years to
almost 7000 years ago>>), that statement implies that his sampling
procedure mixed carbon-bearing substances with two different proveniences:
the loose brown silty crust filling the grooves; and the weathering rind
of the rock at the bottom of those grooves, under the crust.
Even if one leaves aside the issue of the lack of integrity of the mineral
accretions analyzed, it is quite clear that this admitted lack of precision
in sampling, together with the extreme thinness of the mineral accretions,
force us to bring up the question of what exactly it is that Watchman's
samples actually represent (as opposed to what he thinks they represent).
If it was not possible to separate the loose crust from the weathering rind,
was it possible to separate the crust containing the carbon presumably contemporaneous
with its formation from the surficial film of the crust upon which lichen
and other organisms developed? And was it possible to separate the weathering
rind from the unweathered rock itself? The implication of Watchman's explanation
for the graphite problem is that such a separation was not done and is probably
not feasible. In these circumstances, it is quite possible that the <<organics>>
in his samples come essentially from only two sources: the old carbon from
graphite in the rock and its weathering rinds; and the modern carbon from
the living organisms that inhabited the grooves for the last few months
or years. The greater or lesser weight of the latter would determine the
specific <<chronological>> place of each sample in the spread
of dates obtained.
Watchman might of course reply that graphite is a problem only in the case
of the silty brown material found inside the grooves defining the petroglyphs,
not as concerns the silica skins covering the adjacent rock surfaces. His
interpretation of the radiocarbon ages obtained for the latter also implies,
however, that they too were subject to contamination. According to Salema
(1995), the oldest result Watchman obtained for <<organic matter>>
encapsulated in the silica skins covering the rock surfaces adjacent to
the petroglyphs was 1700 BP, and that was, therefore, in the framework of
his assumptions, the maximum age the petroglyphs could have. This conclusion
was reinforced by means of a contextual argument. <<A clue to their
true age came when Watchman learned that the remainder of the brown layer
consisted of silt probably eroded from the hillsides above when farmers
began cultivating. That happened about 1700 years ago -- which he thinks
is the maximum age of the images>> (Fischman 1995). That is, ca. 1700
BP the environmental change brought about by the beginning of agriculture
implied that silica skins ceased to form; inhabitants of the area subsequently
engraved the rock surfaces covered by these silica skins; the grooves were
then filled with a loose silty brown crust made up of material eroded from
the surrounding slopes and containing varying amounts of old charcoal, notably
graphite, responsible for the anomalously old <<minimum ages>>
(Watchman 1995a, 1995b).
Salema (1995) also reports, however, that not all of the silica skins gave
the same radiocarbon age of 1700 BP: most gave younger ages and, in at least
one instance, one such skin from an unengraved rock gave an age of 4300
BP. On the other hand, Watchman believes that these crusts <<do not
take very long to form>>, and he thinks that crust formation in the
walls of a nearby 100 year old quarry is, in both thickness and duration,
a good analogy for the processes that affected the kinds of rocks that were
engraved (Salema 1995). Therefore, the range of dates he obtained for the
silica skins covering those rocks carries a significant implication: if
such skins are rapidly formed and if they ceased to form as a consequence
of an environmental change that took place 1700 years ago (the beginning
of agriculture in the valley), then the samples from them, with ages that
are hundreds or thousands of years apart, have to be <<contaminated>>
as well, some with younger material, some with older material! This, incidentally,
is also admitted by Watchman in a passage where he states that <<ancient
carbon has contaminated the accretions in engravings and on some rock surfaces>>
(Watchman 1995b), and contradicts his other categorical statements that
the silica skins formed before 1700 years ago are uncontaminated: <<graphite
(...) occurs in thin yellow-brown weathering rinds at the base of the silty
brown accretions, but not in the hard, gray to white amorphous silica>>
(Watchman 1995b). In short: the silica skins covering the rock surfaces
adjacent to the petroglyphs, which were supposed to provide the maximum
age for those petroglyphs, also suffer from the same <<contamination>>
problem that led Watchman to disregard the results obtained for the loose
silty brown crust that were supposed to represent their minimum age!
After this, one cannot escape the conclusion that Watchman's explanation
of why some of his radiocarbon results are correct age assessments and others
are not contains too many inconsistencies and leaves too many unanswered
questions to be acceptable. The presence or absence of graphite seems to
be invoked according to the conveniences of the argument but, in an overall
evaluation of the data supplied by him, it would seem that all of the crusts
he sampled suffer from this problem to a greater or lesser extent. In other
words, everything is contaminated! Since it is obviously impossible to quantify
the extent to which this <<contamination>> affected the samples,
and since he cannot exclude that <<contamination>> in the other
direction (that is, by younger carbon) also occurred, it follows that his
<<dates>> are nothing more than chronologically meaningless
expressions of the values attained in the different samples by the ratio
between the two carbon isotopes 12C and 14C.
In any case, even if one accepted that the formation of the silica skins
he analyzed ceased around 1700 BP, that would not necessarily have to be
relevant for the argument regarding the age of the petroglyphs. As discussed
in the previous section, it is perfectly possible that a patina developed
over a rock surface is younger than the petroglyphs found on that same surface.
Watchman might object, however, by saying that if the engravings were already
there when the silica skins began to form, they should also be covered by
such skins (as in Fig. 3). Since that was not
the case (inside the grooves defining the petroglyphs he claims that only
the loose brown silty crust was present), the moment when the silica skins
ceased to accumulate should indeed represent a maximum age for the petroglyphs,
which must have been engraved after those skins formed. At the September
1995 International Rock Art Congress of Turin, however, Watchman stated
that mineral accretions on the analyzed surfaces were only minimally developed,
making it very difficult to obtain adequate samples (and that had also been
the reason why he had suggested that Bednarik be invited to carry out microerosion
dating). It seems fair to infer from this that the skins in question do
not represent extensive and homogeneous covers and that their absence from
the particular engravings Watchman studied may be, therefore, stratigraphically
irrelevant. The basic problem, however, is that, as shown above, such skins
are <<contaminated>> by non-contemporaneous carbon (as Watchman
himself implicitly or explicitly admits), and the moment when they ceased
to form (whether 100, 1000, 10,000 or 100,000 years ago), therefore, cannot
possibly be determined by radiocarbon dating.
On the other hand, Watchman's attribution of a post-1700 BP genesis for
the silty crust filling the grooves can only be interpreted as a minimum
age for the petroglyphs defined by those grooves. If that attribution were
to be accepted, it would only mean that the petroglyphs were older than
the age of the silty crust, not younger. Therefore, if one accepted that
the accumulation of the silty crust is a process resulting from the establishment
of agriculture in the valley, then the petroglyphs would have to be considered
as pre-dating that establishment. All the more so since the fact that he
recognized <<slightly weathered rock surfaces at the base of the silty
accretions>> (Watchman 1995b) implies that the moment of execution
of the petroglyphs and the moment when the silty material began to accumulate
would have to be separated by a significant amount of time!
Leaving this contradiction aside, it should also be stressed that Watchman's
environmental model of the development of mineral accretions in the Côa
valley rock surfaces is based on three contextual arguments that are totally
unsubstantiated:
Two are unverified assumptions -- that the silt in the skin that covers
the grooves eroded from the surrounding slopes (which is likely, but where
are the analyses that prove it?); and that such erosion only took place
as a result of farming (why not before forest development, in late last
glacial times or in the early Holocene?).
The other is an outright invention -- that such slopes were cultivated for
the first time 1700 years ago (how does he know? what kind of research did
he undertake to justify this assertion? how does he explain away the evidence
for a Neolithic -- that is, at least 6000 years old -- settlement of the
area by farmers?).
The unsubstantiated nature of these assumptions is a very important point
in this discussion, because, upon closer scrutiny, it turns out that it
is the argument relating to the moment when agriculture began in the area
that constitutes the essential logical foundation of Watchman's chronology
for the Côa valley art. As a matter of fact, from the point of view
of the dating results, his conclusion is that only the ages obtained for
the silica skins covering the rock surfaces adjacent to the petroglyphs
are acceptable, providing a maximum age for the petroglyphs themselves.
But, since his results for those silica skins cover a wide range of time,
he also has to sort among them those that are <<good>> from
those that are <<bad>> (or <<contaminated>>). And
it is quite clear from the preceding discussion that the good ones are those
in accordance with the 1700 BP date he presumes for the beginning of agriculture,
which he equates with the moment when those skins ceased to develop. In
other words, the maximum age Watchman allows the Côa art to have is
not based on the radiocarbon results, that is, on the direct dating technique
that was supposed to give him superior powers of chronological estimation,
but entirely on the geochemical and archaeological assumptions (that silica
skins ceased to form as a result of agriculture and that agriculture in
the Côa valley only began 1700 years ago) used to either accept (in
some cases) or disregard (in other cases) those radiocarbon results.
In short:
Watchman's theoretical model of crust formation is based on false, or at
least unverified, assumptions, as is the case with his environmental interpretation.
The results obtained for the Côa rocks contradict the theoretical
expectation derived from the model of crust formation.
The lack of precision in the sampling procedure does not allow identification
of the exact microstratigraphic provenience of the carbon present in the
dated samples.
The ratio between 12C and 14C in those samples is chronologically meaningless,
since it consists of a mix, in varying proportions, of: old carbon (graphite)
from the rock itself and its weathering rinds; carbon that is penecontemporaneous
with the formation of the different types of mineral accretions covering
the rock surface and the petroglyphs (organic carbon from dead organisms
encapsulated in those accretions); and recent carbon incorporated through
pedogenetic and anthropic processes in the rock, the weathering rinds, and
the mineral accretions.
The proposed environmental interpretation represents the only basis to sort
out the <<bad>> dates from the <<good>> dates but,
contradictorily, carries, on one hand, the implication that the silica skins
are also contaminated and, on the other, the implication that the petroglyphs
are pre-agriculture, not post-agriculture.
Radiocarbon should never have been used in the first place due to the open
system nature of the unstratified mineral accretions present and the lack
of integrity of the sampled panels.
In these circumstances, it is quite clear that Watchman's <<maximum
age>> cannot be considered, from a scientific point of view, as a
valid critique of the Paleolithic age of the petroglyphs engraved in the
analyzed panels as determined by stylistic criteria. On the other hand,
such a <<maximum age>> is in total contradiction with the results
obtained by Dorn, who also used AMS radiocarbon dating but arrived at a
completely different chronology: that all the engravings are older than
2000 BP, not younger than 1700 BP, as Watchman states.
Esta página foi realizada para o Nestcape Navigator
2.0
Última actualização no dia 1 de Junho 96
siljan@gemini.ci.uc.pt