Introduction, Equipment, Procedures

The first season of fieldwork involving the extraction of cores of Roman hydraulic concrete took place in Italy between 29 July and 13 August 2002 under the direction of the three project co-directors. The coring equipment, generously purchased by CTG Italcementi Group (Bergamo) specifically for use by the ROMACONS project, was supplied by Cordiam S.r.l. (Como). The equipment consisted of a hydraulically driven coring drive (carotatrice modello M60-0) on a hand operated ratchet mount, with a separate gasoline powered hydraulic pump unit (centralina LP11P) connected with 25 m high-pressure hoses with quick release, resealable fittings (Figure 1: coring device at Portus, POR2002.02). The coring drive, designed for mounting on standard construction scaffolding, rotated pipe sections 1.0 m and 0.5 m long and 0.10 m diameter, threaded at to accommodate the diamond drill bit and core-grabber ring at the lower end, and core extension tubes above (Figure 2: adding coring tube at Portus, PTR2002.01). The core-grabber consists of a split, expanding ring steel ring with corrugations on its inside surface that keep the core from falling back out of the coring tubes when they are lifted from the core hole. A hand-held hydraulically driven coring drill (carotatore manuale idraulico modello MAG15) was supplied as well, and used to prepare holes for anchoring bolts that held the scaffolding in position. Both drills incorporate quick-release fittings for a water source designed to cool and lubricate the coring tubes, remove coring debris, and clean the diamond bits. The water source must be pressurized at 2.0-3.0 atm. After examining and packing the equipment at the Italcementi headquarters in Bergamo on 29-30 July, and purchasing further tools, supplies, and expendables, the group drove to Portus, the ancient port of Rome, in two vans on 31 July.

During inspection trips in March of 2001 and February 2002, three coring locations were selected on the north mole of the Claudian harbour at Portus. Care was taken to select points spread out along the portion of the mole that could be accessed without special security permits, where the structure of the mole seemed to be at least several metres thick, and where the upper surface of the structure was relatively level, strong and coherent. Dottoressa Anna Gallina Zevi Soprintendente per I Beni Archeologici di Ostia graciously allowed sampling at all three points.

At each location, a frame was built of scaffolding pipes and clamps and the drill motor and ratchet track mounted in a vertical position over the coring point. Four aluminum leveling legs with hinged base-plates were fixed to the concrete with stainless steel bolts screwed into expansion bolts set into holes drilled with the hand coring device. The entire frame was then carefully adjusted so that the ratchet track was vertical (Figure 3: coring rack at Portus, POR2002.02; Figure 4: coring rack at Portus, PTR2002.02). The mounting frame, which was partially dismantled after each coring session, naturally had to be adapted slightly at the six separate coring locations in order to fit the local situation. Our initial experience at POR 2002.01 made it clear that the stability of the framework was critical to recovery of a coherent core, and that the bolts of the scaffolding clamps and anchors had to be monitored and tightened throughout the coring process. It also became apparent that a flushing water flow of approximately 3.0 l/min at approximately 2 or 3 atm was essential to the proper functioning of the drill bit and coring tubes. This water flows down in the space between the inside of the coring tube and the outside of the core, around the drill bit, then up around the outside of the coring tube to the surface. The colour of the flushing water and the sediment contained in it allowed the operators to determine whether at that moment the hidden drill bit was cutting through mortar or tuff (Figure 5: flushing water with ground-up tuff, POR2002.03).

Once the frame, water supply, and hydraulic power hoses were in place, the drill bit and core-catcher alone were screwed on to the drill motor bit, and an initial cut ca. 0.15 m deep was drilled into the concrete surface (Figure 6: initial hole, POR2002.02). The drill motor bit was then unscrewed from the core-catcher and ratcheted up, leaving the bit and core-grabber in the hole, allowing addition of a 1.0 m long section of coring tube. The heavy threads of the coring tubes and bits were greased before joining them. The sections were screwed together by hand and with careful application of the drill motor, and unscrewed by use of two heavy plumbing wrenches with a 0.15 m reach (Figure 7: unscrewing core tubes at PTR2002.01). Provision of an initial cut of 0.15 m cut guided the descent of the first 1.0 m core tube extension and minimized vibration. When only 0.10 m of the first extension tube remained visible above the concrete surface, the drill was stopped, the drill bit unscrewed from the first coring extension tube, ratcheted up, and a second extension tube added. Depending on friction and the resistance of the concrete, a total of up to six 1.0 m long extension tubes can be driven by the drill. Although coring at POR 2002.01 stopped at a depth of 1.38 m because of excessive vibration and poor alignment of the core tubes, at the five other coring sites a relaxation of back pressure and the appearance of fine sea sand in the flushing water indicated that the coring tubes had penetrated the bottom of the concrete mass.

After the ancient sea floor was reached, the drill was stopped, and the drill and attached core tubes containing the concrete core were ratcheted up high enough to expose the entire top coring tube and the top 0.10 m of the tube below. A collar was then bolted around the top of the lower tube—to prevent it from falling back down the core hole—and the top extension tube was unscrewed from both the drill bit and the lower tube (Figure 8: collar in use at POR2002.02). The drill and but were then ratcheted down to the exposed tube, threaded to it, and the next section extracted. The process was repeated until all extension tubes, the core-grabber and the drill bit were removed. As each extension tube was unscrewed and tipped away from the coring device, the bottom end was capped to prevent the enclosed concrete core from falling out. The tube was then carried to a plastic sheet on which the core was extruded for photography and study (Figure 9: core from PTR2002.01). The cores were 0.09 m in diameter and of varying lengths.

One unexpected result of the coring is that portions of the core consisting entirely of mortar occasionally were ground to powder, apparently when fragments of hard aggregate came loose from their position in the core and were pushed around inside the lower end of the coring tube by the core-grabber ring. This powder then was carried to the surface by the flushing water, leaving a gap in the core recovered (Figure 10: gap in core POR2002.02, -0.87 m to -1.50 m). Careful observation of the depth of the coring tubes when the flushing water seemed particularly thick and creamy white, along with subsequent measurement of gaps in the cores themselves, allowed reconstruction of the thickness of the lost strata.

Once the coring tubes had been removed, the depth of the core hole was measured, and the hole was filled to within 0.15 m of the original surface with clean quarry sand. This fill was compacted with water and by tamping, and, where possible, the original upper 0.10 m of concrete core was replaced, set in a mortar consisting of a 1:4 mixture of lime and pozzolana. Red or grey pozzolana was used according to the mix of the original concrete. Where the original plug disintegrated during coring (as a result of weathering), the plug was made up of aggregate from the area of the core hole, set in pozzolana mortar. Sand and a lean pozzolana mortar mix were used as fill on the advice of E. Gotti, a materials scientist with the Italcementi group, in order to avoid possible stressing of the block by expansion of a plug consisting entirely of modern pozzolana-lime mortar. All coring locations were mapped with a portable Garmin 48 GPS unit and by measurement from nearby permanent structures. Activities and cores were documented with emulsion and digital photography.

Preliminary Analysis of Results

Laboratory analysis of the cores now underway will provide information concerning the origin of the cementing materials and the aggregate used in these harbour installations, the ratio of lime, micro-aggregate and macro-aggregate, and the strength of the resulting concrete. Preliminary, visual analysis already allows some deductions. The North mole of the Claudian harbour basin at Portus is a surprisingly thin structure, and is composed of concrete containing numerous poorly hardened sections (POR 2002.01 to 03). It is possible that the surviving sections of the Claudian mole were in fact a concrete foundation for a more substantial retaining wall and wave breaker built of blocks of stone, now lost to plundering. Alternatively, the upper portions of a concrete mole have simply been lost to weathering, possibly because of the relatively poor quality of the construction materials. The west breakwater of the Trajan harbour (PTR 2002.01) and the Southeast breakwater at Anzio (ANZ 2002.01), in contrast, are substantial structures, composed of a very resistant concrete. There is evidence from both these latter structures that the concrete was poured in formwork inundated with sea water. The core from the quay wall near the entrance to Trajan's harbour basin (PTR 2002.02) shows that this structure is also surprisingly thin, but well built. In addition, the close resemblance of the two samples from the Trajanic harbour suggest that the West breakwater was constructed at the same time as the quay wall, probably as part of the initial, Trajanic construction period.

Thanks and Acknowledgments

This project, which required so much hardware and organization, and so many permissions, could not have taken place without the generous assistance of many individuals. At the CTG Italcementi Group we would like to thank in particular Dott. Luigi Cassar, Central Manager of Research and Development, who showed great faith in the value of this research and facilitated the donation of the crucial coring equipment. We also received great assistance from Dott. Emanuele Gotti (Laboratory Director), Mr. Dario Belotti, Ms. Isabella Mazza, and Mr. Piero Gandini. Dr. Lester Little, Director of the American Academy in Rome, and Dr. Archer Martin, Director of Archaeology were both very helpful in obtaining permissions to work at Portus and Anzio. We are very grateful to Dottoressa Anna Gallina Zevi, Soprintendente, Soprintendenza per i Beni Archeologici di Ostia for granting us a permit to take cores at Portus, and to Dottoressa Morelli, Direttore del Museo delle Navi di Ostia and her staff for facilitating our work there. Thanks are also due to Dottoressa Lidia Paroli, who assisted us in locating suitable sites to sample in the Port of Trajan. Dottoressa Anna Maria Reggiani, Soprintendente, Soprintendenza per i Beni Archeologici del Lazio, kindly gave permission to take the core at Anzio, and Dottoressa Annalisa Zaratinni, Direttore Nucleo Operativo di Archeologia Subaquea, Soprintendenza per i Beni Archeologici del Lazio, gave us crucial last-minute assistance in working out the local arrangements at Anzio.

Description of Core Samples

The North Mole of the Claudian Harbour at Portus.

1) POR 2002.01. The first core sample was taken on 1-2 August from a point located approximately at the midpoint of the North mole, 7.5 m E of the road leading to the main Leonardo da Vinci Airport terminals, 1.9 m from the south face of the mole and 3.7 m from the north face. (33T0271604. UTM 4629777).

The core hole was drilled to ca. 1.38 m, at which point the machinery was stopped because of increased vibration and because the core barrel had become slightly out of alignment with the drill rack. The initial coring effort was hampered by an insufficient water supply, and the resulting vibration apparently affected the alignment of the coring device and frame. The core was retrieved in very broken condition, since much of the mortar had been ground up and washed away, along with some of the tuff aggregate. The final result was one core fragment L 0.11 m, taken from about -1.0 m, along with fragments from above -1.0 m totaling L 0.15 m. Off white mortar and brown to light yellow tuff.

Lower core: 5 pieces tuff aggregate and mortar; ca. 30% mortar (Figure 11). Brown (10YR 5/3) and strong brown (to reddish) (7.5 YR 4/6) tuff aggregate. Mortar is off white with very mixed, pebbly micro-aggregate in grey, brown, red, black. A few small lime nodules. The sorting of micro-aggregate is not uniform, but the mortar seems quite well mixed.

Upper fragments same as lower core.

Evaluation: The intact sample is fairly hard and compact, but much of the block may consist of fairly poor cement, since it disintegrated during sampling. It is possible that the vibration caused by insufficient anchoring and core misalignment exacerbated stress on the mortar.

2) POR.2002.02.. The second core was taken on 2-3 August at a large mass farther west along the North mole. 2.61 m W of a chrome steel survey marker, 10.26 m from N face of mole, 14.27 m E of road to restricted military area. (33T0271508. UTM 4629726)(Figure 3). Testaguzza has proposed that this mass of concrete is the remains of the lighthouse foundation, and thus poured inside the great barge of Caligula (O. Testaguzza, Portus. Rome 1970: 000) No wood from a hull was retrieved with the core, suggesting that his proposal is incorrect.

Three solid core samples were retrieved: L 1.58 m (0.00 to -1.58 m), L 0.40 m (-1.58 to -1.98 m), and L 0.82 m (ca. -2.32 m to -3.14 m) (Figure 12). A section of the core 0.32 m long is missing, probably from level -1.98 to -2.32 m, which discharged a thick solution of white, limey water from the coring flush.

0.00 to -0.20 m. Off white, very limey mortar with rounded minerals, possibly tuff constituents, or possibly sea sand. Bits of grey and brown tuff mixed in. Very poorly mixed, but hard. Aggregate is a fine-grained light brown tuff.

-0.20 to -0.33 m. Mortar and light yellowish brown, very fine-grained tuff (10YR 6/4, dry). Occasional black crystals, and numerous inclusions of slightly redder tuff. Mortar poorly mixed, but very hard. Contains rounded sand grains, bits of tuff and possibly pumice from 1 mm to 10 mm. Large amounts of lime in lumps of various sizes.

-0.33 to -0.40 m. Light brown, very fine grained tuff, as above.

-0.40 to -0.64 m. Very hard, light grey (2.5Y 7/1; could also be Gley 1 7/N) mortar. Contains rounded, coloured particles, possibly sea sand, along with numerous big and small lumps of lime (D 1 mm to 15 mm), grey tuff and pumice, and occasional lumps of brown tuff as above.

-0.64 to -0.73 m. Fine grained tuff aggregate as above.

-0.73 to -0.86 m. Crumbly, off white mortar with many lime lumps, some pozzolana granules or possibly sea sand, gypsum formations.

-0.86 to -1.38 m. Light brown tuff aggregate as above. Mortar off white to light grey as above (-0.40 to -0.64 m). Many large lumps of lime, and some sea sand. Grey to grey green tuff (no Munsell). Occasional fragments of charcoal, and rope or basketry reeds. From -1.00 to -1.38 m, mix is ca. 80% mortar, 20% aggregate.

-1.38 to -1.58 m. Approximately 50:50 mix of mortar (as above) and tuff aggregate (as above).

-1.58 to -1.98 m. Light brown (± 10YR 6/4) tuff and grey mortar. Once piece of very hard light brown (10YR 8/2) sandstone (?). Tuff contains many small scoriae of light yellow brown (10YR 5/4 to 5/3), along with some of grey shades and numerous black crystals. Also a few pieces of aggregate of a very hard, very fine grained brown (10YR 5/3) tuff (?) with numerous fine black crystals. Ca. 50:50 mortar and aggregate. Grey mortar contains numerous lime lumps up to D 0.01 m, lumps of light grey tuff, and a few large fragments of charcoal (up to D 0.015 m). One fragment of a reed or stick. This seems to be a mortar mixed with a light grey pozzolana, tuff, pumice, but no sand. The same mortar is used for the entire block. The last 0.15 m of this section is all mortar; no aggregate.

-1.98 to -2.32 m. A section 0.34 m long is missing from the lower part of the core. Judging from the colour of the discharge from the water flush, and from the measurements of the continuous core samples, the missing section probably consisted of a stratum of loose, limey mortar from a depth between -1.98 and -2.32 m.

-2.32 to -3.14 m. Mortar and aggregate as above.

-3.14 m. The original sea floor was reached; it consists of an extremely fine grey brown sand with numerous black flecks.

Evaluation: A portion of the structure consisting for the most part of good quality concrete, although there may be pockets of softer cementing materials inside. The sand beneath the block suggests a regime of minimal wave action, so it may belong to the period of initial construction activity, when a lagoon was walled in and transformed into the Claudian harbour.

3) POR.2002.03. The third core was taken on 4 August at a low section near the east end of the visible portion of the North mole, immediately N of Museo delle Navi. 1.97 m  m W of chrome survey marker, 3.84 m from the south face of the mole and 1.29 m from the north face. (33T0271688. UTM4629800).

Surface to a depth of -1.56 m. Solid core L 0.36 m (two pieces). The concrete was very soft and easily damaged. Most of the mortar was ground up during the coring process and discharged along with the flushing water. The exact depth of the remaining solid core could not be determined, but to judge from the colour of the flushing water, it may originate from the upper part of the block. (Figure 13).

0.00 to -1.56 m. Crumbly, light brownish grey (2.5Y 6/2) mortar consisting almost entirely of tuff lumps up to 0.008 m D, grey, yellow, green, white in colour. Well mixed mortar containing a high proportion of small, rounded particles resembling sea sand, but most of it softer than quartz and possibly deriving from the pozzolana. Only a few small lime lumps, and it appears that there is little lime in the mix. It is likely that the soft, limey portions of the structure were ground up during the coring process. The tuff aggregate is a strong brown (7.5YR 5.6) with slightly redder inclusions and grey inclusions; quite hard.

-1.56 m. Sea sand; coarse grained, off white with numerous heavy black particles.

Evaluation. This portion of the mole is thin and very poorly constructed. This whole stretch of thin concrete wall laid directly on sand may in fact be the footings for a masonry wall meant to form the quay or retaining wall to hold back the backfill as the harbour basin was excavated. The structure seems undermined at several points, and the heavy, coarse sand suggests a regime involving wave action. Either the sand was introduced as the harbour breakwaters disintegrated and the wave action undermined this wall, or the wall was built after the harbour basin had been in use for some time.

Southeast Mole of the Neronian Harbour at Anzio.

4) ANZ.2002.01. The fourth sample was taken on 6 August from the middle of the block at the base of the Southeast Mole, adjacent to a parking lot. 10.99 m from the northwest corner of the block; 12.09 m from the northeast corner; 14.09 m from the southwest corner; 13.92 m from the southeast corner; 3.87 m from the west face; 3.56 m from east face. (33T0302127. UTM 4590670). (Figure 14: view of rack and coring device).

Drilled to ca. 3.10 m. First stage of coring taken to -2.55 m, at which point the water stopped flowing through the tube (because of the build-up of sea sand). Core material removed, and drilling continued to 3.10 m. It is likely that approximately 0.25 m of limey pozzolana at the bottom of the core was lost to the drilling activity. The last 0.60 m consisted of a grey-greenish sea sand. This deposit apparently is quite compact, since the hole through the sand remained accessible to a tape measure even after the coring tube was withdrawn, and despite being filled with water by the flushing system.

0.00 to -0.27 m. Olive brown (moist, 2.5Y 4/4; light yellowish brown when dry, 2.5Y 6/4) tuff aggregate set in a light olive brown (2.5Y 5/4) mortar. The mortar is well mixed and very granular and contains numerous smaller chunks of yellow (2.5Y 7/6) fibrous tuff or pumice. The smaller grains are clear and coloured, rounded grains, possibly sea sand, but more likely grains of pozzolana hydration products. Occasional lime nodules.

-0.27 to -1.90 m. At -0.27, an abrupt change to a light grey (10YR 7/1) mortar with numerous chunks of brown tuff aggregate as above. Below this point the concrete mix is very uniform down to the puddling of the initial pour at -1.90 m (Figure 15: view of core). The mortar contains many lumps of slightly greyer aggregate up to D 0.01 m, along with numerous (but not excessive) lime nodules D 0.003 m to 0.01 m. The mortar is very well mixed and uniform down to -1.90 m. There are two more loosely packed areas (-0.90 m, -1.80 m), but the general impression is that of a very carefully mixed and well-laid material. The ratio of mortar to aggregate seems to be about 60:40.

-1.90 to -1.94 m. The tuff aggregate is absent from the mixture below -1.90 m. The builders apparently dumped a mix of lime and pozzolana sand into the formwork, which was at least partly filled with water, and the coarser materials sank to the sea floor (at ca. -2.25 m) while the lighter materials rose to the surface of the mix. As a result, the cementing materials at -1.90 to -1.94 m consist of layered deposit of extremely fine lime particles, greyish white in colour, with a distinct upper boundary separating it from the mortar mix above.

-1.94 to -1.99 m. The mortar mix is of a slightly more greenish hue, and consists of a very fine sand.

-1.99 to -2.04 m. Coarse greenish-grey pozzolana grains, with much lime (Figure 16).

-2.04 to ca. -2.25 m. A loose mix of greenish grey (Gley 1 5/10Y) pozzolana and poorly mixed lime, retrieved from the tube in crushed form and in lumps up to D 0.06. Much of this deposit seems to have been ground up by the coring tube and discharged with the cooling water, so the thickness of the deposit is uncertain. It is unlikely, however, to have been much more than 0.30 m.

-2.25 to -3.10 m. Coarse, greenish grey sea sand with numerous black grains.

Evaluation. There seem to be three series of pours in this core. The first one was a pozzolana-lime mix dumped without aggregate (unless the core sample is incorrect) directly on the sea floor. Much of this layer separated in the water and did not form a good cementing substance. The second pour was carefully mixed, and poured or laid in a very consistent layer 1.63 m thick. The third pour, which may well have followed immediately upon the second, made use of different tuff additives to the mortar, giving it a different colour. This upper layer, now in contact with the atmosphere, also seems softer.

Harbour works of the Port of Trajan at Portus

5) PTR.2002.01. The core was taken on 8 August toward the north end of the (possibly Late Imperial), west entrance mole. 61.05 m from porch of custodian's house, at bearing of 142 degrees, and 2.0 m from the east face of the mole; 46.6 m east of the boundary fence. (33T0271623. UTM 4628837) (Figure 2).

Drilled to depth of 2.43 m, 0.20 m into the sea sand. The core length is 2.23 m. Core is quite uniform, consisting of off-white, well mixed mortar, about 50:50 mortar and aggregate. The light brown to dark brown tuff aggregate is uniformly sized and spaced. Occasional brick fragments, and one "leveling course" of brick at -1.15 m.

0.0 to -0.21 m. Reddish brown, speckled mortar, with heavy admixture of red, yellow, green, and grey pozzolana granules (to D 0.005 m), and a few lime nodules. Several fragments of brick. Probably the same as the mortar deeper in the block, but weathered. The grains seem more separated, and the mortar softer than that deeper in the block.

-0.21 to -0.27 m. Very hard, very fine-grained, yellowish brown (10TR 5/4) tuff aggregate with occasional black specks.

-0.27 to -1.10 m. Very uniform pour of hard, well-mixed, off-white (10YR 8/1) mortar with heavy admixture of grey pozzolana particles and occasional lime nodules. Fragments of light red brown (2.5YR 6/3) and very pale brown (-light yellow") (10YR 7/3) brick (Th 0.022 m, 0.035 m). Light grey (2.5Y 7/2) to dark brown (10YR 4.3) tuff aggregate, some of it speckled with coloured scoriae.

-1.10 to -1.12 m. A section of a light red brown (2.5YR 6/3) brick occupies the entire core. This brick course can be seen on the face of the concrete mole.

-1.12 to -2.0 m. Same concrete mix as above (Figure 17).

-2.0 to -2.23 m. Mortar is the same as above, but less well mixed, and softer, with seams between many of the granules. Aggregate at base of pour, above the sea sand consists of two large sherds of terracotta, possibly from amphorae.

-2.23 to -2.43 m. Fine, light brown sea sand.

Evaluation. A uniform, well prepared pour, with at least one brick "leveling course," laid on a fine, possibly lagoonal sand.

6) PTR.2002.02. Core taken on 9 August at the west edge of N/S quay wall in front of "Severan warehouses", just north of the entrance to the hexagonal basin, in an excavated pit surrounded by a wooden fence. (33T0272140. UTM 4628884. (Figure 4).

Drilled to depth of 1.67 m, ca. 0.10 m into the sea sand. The core length was 1.65 m. Core is quite uniform and similar in materials to PTR 2002.01. It consists of hard, well-mixed, slightly yellowish grey mortar containing a high proportion of grey, yellow, red, and brown pozzolana granules up to D 0.005 m. Reddish brown, yellow brown, and greenish tuff aggregate. The aggregate is uniformly sized and spaced. (Figure 18)

0.00 to -0.20 m. Light yellowish brown (ca. 2.5Y 6/3) mortar with large pozzolana granules and occasional lime nodules. Same as the mortar farther down in the block, but looser, with open seams, as if weathered. Aggregate consists of a very speckled red to brown tuff that appears to be almost a conglomerate, and a yellowish red (5YR 4/6) speckled tuff.

-0.20 to -1.40 m. The mortar is the same as that above, but unweathered, and thus a greyish white (10YR 8/1, 9/-). Well mixed mortar, with much darker grey pozzolana granules to D 0.005 m. Tuff aggregate as above. Approximately equal amounts of aggregate and mortar.

-1.40 to -1.65 m. A layer of chalky, light greenish-grey (Gley 1 8/10Y) lime (Th 0.05 m) above sorted granules of light greenish grey (Gley 1 7/1) pozzolana, the heavier particles deeper than the lighter. Clearly a first pour into an inundated form, where the cementing mix was sorted by the water. Two pieces of yellowish red tuff at the bottom of the pour, along with lumps of tuff ground up by the coring device. (Figure 19).

-1.65 to -1.85 m Relatively coarse, dark grey sea sand with numerous black particles.

Evaluation: A well-mixed pour, at least the first portion of which was laid in an inundated form on sand subject to wave action.