Cochineros is a petroglyph site on the right bank of the Rio Mala, at an elevation of 570 metres above sea level. It is a long broad area of elevated natural terrace running beside the river – five hundred metres long and up to sixty metres wide – some four metres above the winter water level.
There are hundreds of rocks and boulders on the natural terrace, of which twenty or more are marked with images. They vary from partially buried stones with just one face visible, to giant boulders which have been decorated on every face. Three marked stones now lying on the river bed may once have rested on the terrace, as the river is moving westward and eroding the land. The largest boulder is four metres wide and ten metres long, with a surface area of forty square metres containing images. This is the size of a small building – other boulders are as large as vans and cars.
The site is 30 kilometres inland from the coast and 45 kilometres downstream from Huarochiri. There are several other known petroglyph sites in the valley. Calango, a town 12 kilometres downstream, is noted as the site of a petroglyph investigated by Antonio de La Calancha in 1622. He in turn visited because he had already heard stories about the stone, and he collected and wrote down both the oral histories and religious speculations about the marks on the stone. Retama, a site with one giant boulder and one smaller stone, is one kilometre downstream from Cochineros, on the opposite bank of the river, with a view down onto the site. I have found two individual petroglyphs close to the road up from Calango. Erasmo Quispe of the Municipality in Calango has found and photographed stones by the river at Minay, some five kilometres upstream, and Checas, 13 km upstream. These last two riverside sites exhibit many similar markings – camelids, tumi shapes, groups of dots – to the Cochineros stones. All the riverside stones used as canvases are of the same, hard, black-grey diorite. The Calango stone is quite different, with a polished grey-white surface, with the markings also smoothed and polished, except where the Spanish defaced them. Three minor stones at Cochineros, together with one in the centre of the village of Minay, are brown stones with a reddish surface patina through which the petroglyphs have been cut.
The stones were visited by Antonio Nuñez Jimenez in or around 1976. [1]
He published 36 pages of drawings and photographs in his “Petroglifos del Peru” and what he records is meticulous and almost wholly accurate. He numbers his stones up to twelve, although no images are recorded for some stones, and some minor stones are missed. However his published work misses some major panels, perhaps 25 percent of the artwork at the site.
The most interesting omission is a darker layer of images on several panels which is only apparent under certain lighting conditions.
Henry Tantalean described the site in 2010. [2] He began to tabulate and categorise the images, and measure the stones, but the work is incomplete. The site is large, with a steep drop to the river bed meaning that several panels are difficult to access or view. He concluded that the many tumis at the site, together with Inca sites in the valley, indicated that the majority of images dated to the Inca period. He also did not notice the dark images.
Photographs of images and panels are included in a photographic record of petroglyph sites by Guffroy in Imágenes y Paisajes Rupestres del Perú, 2009, together with a brief summary of the site . [3]
In the terms of Guffroy’s discussion of possible reasons for the locations of rock art sites, Cochineros is a theatrical site, which can be viewed in its entirety from the far side of the river; it is a site where thousands of people could gather, with a flat terrace of 16000 square metres; it is on a route and a probable roadway between two major religious sites, Pachacamac and Paria Caca; it is more or less half way between the coast and the highlands, in the Chaupi Yunga, the fertile zone; it may have been a prime coca growing area, in a warm fertile valley – Antonia de Calancha records that coca was grown “twelve or fourteen cuadras” upriver from Calango; it is besides a river which flows throughout the year; and it is a kilometre downstream from a major pre-hispanic settlement – though in truth there are the remains of villages every few kilometres along the length of the valley from the coast to 50 km inland.
A study of the valley by Johnny Taira recorded other petroglyphs at Minay and Checta and suggested that a pre-hispanic road from the coast to Paria Caca was evident in the valley and most likely passed the Retama and Chincheros sites. [4]
The site is difficult to access as the river can only be crossed on foot for a few months of the year. The images are very well preserved with little vandalism. Nuñez numbered his drawings in reference to stones 1 to 12, though no drawings are recorded for stones 7 and 9. Tantalean believed there were 14 stones. He began to count and categorize the images on eight of the stones, including the five most important, and recorded over three hundred images. The apple growing farmer who owns the land and works amongst the stones says there are thirteen. I have recorded twenty stones with images. However, the majority of the individual images, as well as the largest panels, are found on just five stones.
Tantalean records these five stones as all having one side greater than three metres in length. The plain words “stones” or “boulders” do not do justice to these magnificent rock outcrops. Piedra 11 (following Nuñez’ numbering system) was measured by Henry Tantalean as 3.75 x 10.35 metres (maximum length x maximum width) and the five most important stones have a combined surface area with drawings of one hundred and fifty square metres. Although the number of stones does not compare with sites such as Checta, Toros Muertos or Miculla, where hundreds of small rocks are spread over an area, the number of images drawn here (700 to one thousand) is comparable.
It is evident at first sight on some panels that there are images of differing brightness. This paper presents an attempt to use this varying brightness to rank the drawings by age, and then to calibrate the ranking to get an absolute age estimate. The estimates are supported by comparing images with those found on Late Intermediate and Middle Horizon textiles and ceramics.
It is possible to distinguish brighter, whiter drawings and duller greyer drawings on many of the rock panels. This is more evident to the naked eye on some panels – generally vertical or nearly vertical panels facing south or east – than on others, such as flat or north facing panels. Where one image has been drawn over another older image here, the newer image is brighter.
If the brightness can be consistently measured, it can be used for some quantitative assessment. If it also indicates age, then the drawings can be arranged in a age-related series. If some points in this series can be calibrated, and the mechanism of the brightness change is known so as to justify why it would be, for example, linear with time, then we would have a method of dating.
A panel with clearly differentiated bright and dark images was chosen. The panel (Figure 1) is two metres high and three metres wide at the base, rising to an apex. Jimenez records some thirty individual images on the panel in his Fig 1488.
To see whether the changes in brightness or the patina could be measured quantitatively, photographs of the southern facing panel of Piedra 11 were taken at different angles.
Images were initially taken with a digital camera at an angle to the rock face to avoid directly reflected sunlight, at 1.19 pm and 1.20 pm on 22 August 2016, at angles of approximately 30 and 45 degrees to the perpendicular to the panel.
The brightness of different images on the panel was measured using open source image analysis software (Fiji or ImageJ). The selected images were the Old Llama, Tupus 1, 2 and 3, Lake Deities left and right, the Dot Track and the Heart.
As expected, the measurements agreed with the naked eye, and the images can be ranked from brightest to darkest – Heart, Dot Track, Tupus, Lake Deities and Old Llama.
The absolute brightness values were different for the two photographs – but when brightness was plotted against a horizontal scale that could relate to time, both gave similar straight line graphs. That is to say, their relative brightness as measured by the software was consistent.
When a camera creates Jpeg images, it modifies the raw digital data. Typically it stretches the distribution for the lighter parts of the photograph, to compensate for the human eye’s reduced sensitivity for brighter images.
To eliminate this, the same rock panel was again photographed with a more advanced camera that would provide unmodified Raw images ( .cr2 files). These were taken on 30 June 2018 at 12.03 pm (4606.cr2) and on 21 July 2018 at 1.45 pm (5066.cr2, 5067.cr2, 5068.cr2). This also provided a check on to what extent the results were dependent on the angle of the sun, as I now had images both at different times of day and different times of the year.
Once again, whilst the absolute values of brightness varied, their relative values did not. That is, the measured brightness of the images analysed still displayed a straight line relationship. As long as direct sunlight reflection was avoided, the results were not greatly affected by sun position, angle of the camera towards the panel, or the data compression in saving as a jpeg.
So, with care, it was possible to make a consistent quantitative measurement of selected images on the rock panel. The next step was to relate this to age of the images.
There are several clear over-drawings on the panel. For example, three tupus, or shawl pins, have been drawn over a less distinct camelid, the Old Llama, and a bird-camelid morph has been drawn over a spiral. From the overdrawing alone, it is possible to identify different stages in the use of the site. The upper images on the panel, such as the tupus, are brighter than the images over which they are drawn. This is observable across the whole site.
As the upper images are consistently brighter, this suggests the darkening process is time related. This would mean that the brightness sequence Heart-Dot Track-Tupus-Lake Deities-Old Llama also implies a time sequence from most recent to oldest. This means we can stack them up like strata and say one is older than the next.
To determine the absolute age of the images I need to calibrate points on the line from images of known date. I also want a model for the relationship between brightness and age – ideally, I want to know the mechanism of the darkening process. This would enable me to justify, for example, that the relationship between brightness and age is linear, and the points can be fitted to a straight line.
Modelling the relationship between image brightness and age
I propose that where the images on a panel at the site show varying brightness, it is caused by the growth of a thin film over the image. When the image is freshly made is stands out as bright white against the dark rock, and as the dark film increases in thickness, the white becomes darker.
Localised darkening can be seen on some rock surfaces at Cochineros. Piedra 1 (Figure 2) has a long vertical panel facing the river. At its foot is a white horizontal band above the present ground level, which is broken in the centre by a dark vertical band which descends from the upper surface of the rock.
A kidney-shaped figure which overlaps the band is clearly darkened within the band area. These observations are consistent with the formation of a dark film on the rock surface, gradually obscuring the drawings.
The appearance of the rock surface above and below the kidney supports the proposal of film formation – it has a darker, shinier surface compared to the surrounding rock. It appears to be an area where moisture runs down the face of the rock. This might be from rainfall, moisture or spray from the river when it is at is highest, or condensation from the warm daytime air on the rocks which have cooled overnight.
There are several films which form on rock surfaces in a desert environment which have been widely studied. A dark black film with a high manganese content was first observed by Humboldt on the banks of the Orinoco, and is commonly seen as a coating on rock art in New Mexico.
Fragments of the glossy black rock surface at Cochineros are weakly ferromagnetic, which might indicate iron or manganese components.
A reddish brown coating appears at other sites (in Peru, petroglyph sites at Checta and Miculla have this reddish coating, as do three rocks at Cochineros).
This film is often called rock varnish and has been documented in many desert regions. The films have been studied in detail and have also been used, in various ways and with mixed success, to date petroglyphs.
The rate of growth of this “rock varnish” has been estimated to be 1 -40 micrometres or microns per thousand years, as an average over 10-20,000 year periods, based on measurements at different sites in different environments.
If a 75 micron coating – equivalent to three coats of paint – were required to entirely hide the whiteness of the drawn image, then these growth rates would cover the drawings in two thousand years, at the maximum rate of growth.
Estimates of the rate of rock varnish growth are questionable and in any case are considered to vary from place to place and even from one side of a boulder to another. I simply note that the growth rates above are compatible with the date estimates here.
In order to use the change of brightness due to film formation to support dating estimates for different images, the film formation must be consistent across a panel. This is clearly not the case in Figure 2 above. Here and on other stones, darkening appears to take place more strongly where there is a sloping channel which could collect moisture run-off. So for consistent film formation we want a flat panel, without run off from an upper surface. The south panel of Piedra 11 (Figure 1) largely satisfies these conditions in its centre.
Consistency of brightness within individual images on this panel, suggests that the film formation is uniform across the width and height of the image.
In particular, the dot track, which runs for a length of two metres from close to the top to close to the bottom of the panel, appears to have a consistent brightness until the lowest seven dots, which are also smaller. Quantitative measurements confirm this.
Similarly the heart extends across 60 cm horizontally and vertically, and appears consistently bright. The white band at the foot of the panel is unbroken, which suggests there are no vertical areas with differential film formation.
The top and the shoulder of the rock shows signs of differential brightness, and there are some ridges lower down the surface which may affect the film growth. Images here have not been considered suitable for measurement.
Calibrating the relationship between age and brightness
We have a powerful aid to estimating the ages of drawings on this panel, since the brightness measurements can be calibrated with a modern image, the heart which is dated 1964.
There is a minor complication here: it seems unlikely that “1964” is the date of the drawing, since it is not shown on Antonio Nuñez’ Fig 1488 sketch of the rock surface.
Nuñez was Cultural Ambassador to Peru 1972-1977 and published his four volume work on Petroglyphs of Peru in 1986. So the heart would appear to have been drawn after 1972. The heart is seen on a photograph by Tantalean which was published in 2010, though that photograph could have been taken a few years earlier.
It could be that JMJR met his partner or got married in 1964, and the engraving was made twenty or thirty years later. For the purposes of calibration, these uncertainties in date are not significant, compared to the uncertainty in pixel brightness values. I have taken 1995, or 24 years Before Present, as the calibration date for the heart.
A second calibration point is provided by the assumption that writing on the rocks could not have continued long after the Spanish invasion. The Spanish reached Pachacamac in 1533. They founded Lima in 1535. In 1571 the Reducciones began, whereby Peruvians were forced to move to new settlements under Spanish control, Calango being one such. By 1586 Corregidor Davila Briceño had moved the people 0f 200 Yauyos villages into 39 Spanish centres. In 1608 the Extirpacion de los Idolatrias began its visits throughout the Huarochiri region, destroying sacred sites, and intimidating those who practised the indigenous religion. In this context, it seems unlikely that people could have continued for long to visit their holy places and leave their marks upon the rocks.
Using the heart calibration date of 1995 plus or minus twenty years, 24 years BP, and the next brightest historic image as the dot track with a calibration date of 1580 plus or minus twenty years, 440 years BP, we get a consistent variation in brightness using jpeg images at slightly different angles to the rock face, and RAW pixel data (without adjustment) from photographs taken at a different date and from different angles.
The calibrated points on a linear graph suggest date estimates which are consistent across two jpg images taken at different angles to the rock face on the same day, and four .cr2 RAW images, three taken on one day covering different areas of the rock panel, and the fourth taken three weeks later. 215.jpg was taken at 1.21 pm on 22 August 2016, 219.jpg at a slightly different angle on the same day, 4606.cr2 taken at 12 pm on 30 June 2018, and 5066.cr2, 5067.cr2 and 5068.cr2 taken on 1.45 pm on 21 July 2018.
The data provides a time sequence for the images on the rock panel. It also gives age estimates for different images that span over one thousand years.The oldest figures are put at 1500 BP. The range of values for the brightness of our calibrated points leads to a an uncertainty of plus or minus several hundred years for each of these dates. More advanced statistical analysis may provide a narrower estimate.
Linear v Exponential Relationship
The linear fit simply assumes that if the image brightness falls, for example, by 20, from 220 to 200, in 400 years, then it will fall by 100, to 120, in 2000 years. In fact if the darkening is caused by a film which grows at a constant rate, the relationship would be exponential. That means if it falls by 10%, from 220 to 200, in 400 years, it will fall by 10% plus 10% plus 10% plus 10% in 2000 years. This means it falls to 90% to 81% to 74% to 62.5% (or to 137 pixel value). This makes little difference to the ages of the brighter images such as tupus, but suggests the darkest figures could be a few hundred years older. As an example, the exponential plot for last photograph analysed above, 4606.cr2, is shown below.
The exponential fit depends on the assumption that the film forms at a constant rate. If its formation is dependent solely climatic conditions – temperature, humidity, wind – then these are not expected changed significantly over the relative short timescale of several thousand years.
An ability to fix the date of any images by an independent method, such as by comparison with known iconography from ceramics or textiles, would help confirm the calibration. Happily this is possible. Many drawings on other panels recall Chancay-style textile or Late Horizon images.
There are engravings of sea-birds which are similar to Chancay period central coast textile imagery and also reflect engravings on a rock at Lunahuana, in the Cañete river valley to the south. Three fabulous birds, which resemble Chancay textile imagery found at Pachacamac, are drawn on the facing panels of three stones on the eastern, river bank, border of the site. There is a pair of felines close to ground level on the west side of the southern most stone, drawn in a style similar to Chancay woven textile designs.
These drawings are not on the dated panel, and so the comparison is only qualitative, but they are clearly visible bright images, and they can be assigned stylistically to the Late Intermediate or 1000 to 500 years BP.
Some images at the site may represent llama sacrifice, which could be related to the worship of Paria Caca described in the Huarochiri Manuscript, written in and around 1622, the start of the extirpation. The start of Paria Caca’s presence in the mountains and valley is not known, but a clue comes from the reference by Extirpador Acosta to one of their founding fathers, Tutay Quiri, whose mummified remains, taken to Lima to be publicly burnt as a gesture of destructive contempt for everything Peruvian, were estimated as being 800 years old, that is from 1200 BP. This Paria Caca period then could be from 1200 to 400 BP. Elsewhere, I argue that some images on this panel relate specifically to Paria Caca. These include the lake deities (at around 1000 BP) up to and including the dot trail at 44o BP.
More measurements under controlled lighting conditions might allow selected images – the tupus, and the tumi images with the double circled handle, for example – to be used, with care, as dating calibrators across the site. This might allow cross-calibration from panels with different orientation, to enable the age of images on several other panels and rocks to be estimated. However the possibility of varying rates of film formation from panel to panel and across panels is challenging.
At the opposite end of the brightness and time scale, there are dark images – those that do not appear as white marks on the black stone surface.
The dark images are most evident when the sun at its northern limits close to June solstice shines on the southern slanting face of Piedra 12. Images that are barely visible under normal illumination (Figure 10) stand out sharply as dark against the shining rock surface (Figure 11). A translation of the photograph to adjust the vertical dimensions shows even more clearly the abstract nature of these images (Figure 12). On other rock panels, artificial illumination at night would help to reveal them.
The central panel in Figures 11, 12 has 30 abstract images including parallel lines, curves and dots, no two of which are the same. In the same area, the white images consist of four curving tailed quadrupeds (the same image repeated four times), some swallow-like birds, a man with a line of four llamas, and another human figure with one llama, very poorly drawn.
The majority of the dark images seen above are abstract and geometrical. They are varied and individual with little or no repetition. They are deeply inscribed, and evident from the surface depression or change of texture rather than colour change. On the estimated time scales above, these may date to 1500 to 2000 years BP or earlier.
Once the rock varnish layer has completely opaqued the white, there will be no further change in colour. Some other analysis, for example of surface texture, might reveal that these images go back further in time.
Dark camelid images are drawn with four legs and as if running, and in ones and twos, quite unlike the highly stylised rectilinear llamas which appear white on black on the same panel. They could represent wild camelids depicted by hunters rather than those organised lines drawn by a culture using them for transport and sacrifice.
The proposed dates for the dark geometric motifs align with Lima culture, whose ceramic styles were also geometric, in contrast with the brighter Late Intermediate Chimu/Chancay/Ychsma figurative motifs of camelids, birds and felines at Cochineros. A comparison of the dark motifs with those seen on regional ceramics and textiles of the Early Intermediate and Middle Horizon may provide further insight into the site.
These drawings could therefore be contemporary with Lima culture, to the north, in the Lurin, Rimac and Chillon valleys and Cerro d’Oro to the south in the Cañete valley.
A closer investigation of other rocks at Cochineros reveals that several of them have such dark images, though they are not immediately obvious. A powerful parallel light source used at night would help observe them clearly.
It would appear that the site has been used as a drawing board for close to two thousand years, with an older abstract imagery being marked over by figurative images of birds, llamas, fish, tupus and tumis. Several phases of drawing are evident, and there is scope for much more detailed study of this complex site.
A tentative sequence and chronology for Checta, Peru, Gori-Tumi Echevarría López, 2011, Rock Art Research Vol. 28, No 2, pp. 211-224
[1] Petroglifos del Perú: panorama mundial del arte rupestre, Antonio Núñez Jiménez, Editorial Científico-Técnica, 1986 – 1307 Seiten
[2] Los petroglifos del Valle de Mala, costa centro sur peruana: una explicación materialista- histórica, Henry Tantaleán henrytantalean@yahoo.es Universidad Autónoma de Barcelona, España /Universidad Nacional Mayor de San Marcos.
[3] Images and scenery of Peruvian rock art, J. Guffroy, Institut de Recherche pour le Developpement, 2009 – 307 Seiten
[4] ARQUEOLOGIA DEL VALLE MEDIO DE MALA EN EPOCAS TARDIAS, CAÑETE- LIMA2
Johnny Taira C.1