Our goals this season in M23 were to interpret the relationship between L23 and the ancient road, as well as to establish an accurate chronology for the area. We focused the majority of our attention on sinking a small 225-cm x 150-cm probe into the ancient road surface. As we hoped, our excavation of this probe provided us with a clear stratigraphic cross-section of the road. From this cross-section it is possible to reconstruct a reliable building history for the ancient road and the associated structure in L23. It is also important to note that the phasing of the road is largely indicative of the phasing of the site as whole, and may prove useful in dating the temple.

We began excavating the final layer of topsoil, locus [LM23001], on 31-5 at an elevation of 182.37. This layer consisted mainly of water washed tumble: cobbles, gravel, highly worn pottery, tesserae, glass, and three roofing tiles. Loci [LM23002] and [LL23001] are nearly identical in composition and content. Thus we gleaned little new information from this locus.

On 1-6 we continued down into locus [LM23003]. [LM23003] is identical to locus [LM23008] in the western half of M23, and it is clear that these two loci represent one surface. This surface was made of moderately packed gravel/lime slurry, and probably served as a road. Because [LM23003] is so near the surface, parts of the road have eroded as water washed down the hill over the centuries, badly damaging the strata. The pottery retrieved from [LM23003] spanned from the 4th to the 6th century, suggesting that the construction of the road dated between the 5th and 6th centuries. It is possible that the road was colonnaded in this period since we found approximately 5 roofing tiles and a roofing nail, or it is also possible that these artifacts may have fallen off the structure in L23.

We next encountered another gravel road surface at an elevation of 182.14. We designated this road surface locus [LM23009]. The elevation of the road is very significant because it is nearly the same elevation as the water channel cappers [182.17] in the western probe, suggesting that the construction of the water channel and road are related. The pottery recovered from locus [LM23009] was highly worn, and likely contaminated considering the pottery spanned from the 3rd to the 6th century. The pottery from locus [LM23010], however, was much cleaner, spanning only between the 3rd and 4th centuries. Locus [LM23010] was a compact layer of soil located directly beneath the road surface in [LM23009], which probably functioned as bedding for the road. Because this locus was sealed beneath the road surface, the pottery is much more reliable for dating. (It is important to note that we found a large "city coin" [coin #2] while sifting material from [LM23010], and a golden earring. The coin and the earring may assist in the confirming the date of the locus.) The pottery evidence from [LM23010] suggests that the construction of the road and the water channel date to the 4th century, probably the late 4th century following the earthquake in 363. In this scenario, the earthquake probably disrupted a pre-existing Roman water conduit, and the Byzantines rushed to install a new water pipe in the structure in L23. In their haste to fix the damage the Byzantines simply laid the new channel over the existing road, and removed part of the stylobate. They wanted to continue using the road, so they laid fill around the channel and built a new road over the top. There is also evidence of this late 4th century building project in L23. At an elevation of 181.14 existed a hard-packed floor laid over rubble, also full or late Roman pottery. The Byzantines probably put down this floor after ripping up the original floor to install the pipe.

Not surprisingly, the concrete bedding for the water channel cross-cuts [LM23011], which lies beneath the fill for the 4th century road beginning at an elevation of 181.91. This locus contained two distinct surfaces. The first was a thin layer of cement, plaster, travertine, and gravel packed approximately 5 cms thick that had been packed into an extremely hard road surface. Sealed under this thin layer of cement lay the leveling fill for the road. Interestingly this fill consisted primarily of worked basalt chippings. [LM23011] is characterized by a predominance of these chippings, which range in size from several centimeters to nearly 20 centimeters. Jack believes these chips were the waste products from the shaping of basalt building stones by the Byzantines. Assumedly, the Byzantines used these waste products to level out the foundation for the road surface. The vast majority of pottery obtained from [LM23011] dates from the 3rd century. One piece of Havarit ware possessed a range of use from the 3rd-5th century. I am confident that [LM23012] is a clean locus and the pottery evidence can be used to reliably date the road to the 3rd century. Because the water channel interrupts [LM23011], the channel certainly must have been built later than the 3rd century.

On 6-13 at an elevation of 181.82 in the south and 181.48 in the north we uncovered another thin layer of travertine cement beneath the basalt layer. We designated this strata locus [LM23012]. [LM23012] was also approximately 5 cms thick, and extremely well preserved. Not much pottery was retrieved from this thin road layer, but what we did find spanned from the 1st to the 3rd century. This span is based only off 8 diagnostic sherds, seven from one 2nd-3rd century cook pot. The pottery evidence from the foundation fill for this road, locus [LM23013], only complicates the dating of this surface. [LM23013] began at an elevation of 181.75 in the south and 181.45 in the north, and consisted of moderately compacted soil mixed with gravel, cobbles, and boulders. Also mixed into the matrix was a huge amount of highly fragmentary pottery sherds. Unfortunately, these sherds ranged in dating from the 1st-6th century. (It should be noted that only 1 shard of a 6th century oil lamp was found to give such a high range.) The majority of the diagnostic sherds fell into two categories: 1st and 2nd century ESA ware (40), and 1st-3rd century disk lamp fragments (10). Although, Nick did identify 12 separate pieces of 3rd-5th material. Using this pottery to date the locus will be very tricky. First of all, there is the problem of contamination. It is possible that material from the water channel bedding which cross-cuts [LM23013] could have contaminated the buckets. (Even though this is unlikely considering we obtained similar pottery readings from [LM23015] away from the channel.) There is always also the possibility of contamination from the balks as well. Besides contamination, Andrea Berlin pointed out that what Nick has identified as 3rd century pottery may have actually extended back into the 2nd century since no reliable typologies exist for 2nd century pottery in this area. Due to these concerns I think the road could be tentatively dated either to the late 2nd or early third century. I favor the late 2nd century date based on the predominance of 1st and 2nd century pottery in the locus, while Nick favors the early third century date based on the presence of 3rd century sherds. It is also my opinion that the structure in L23 was built at the same time as this road. We did discover one small, yet readable coin from this locus that may help to narrow down the dating after it is read.

Because we feared contamination from the water channel, we opened decided to sink a mini-probe safely away from the channel. This probe measured 150 cms x 125 cms. We designated the first locus in probe [LM23015] beginning at an elevation of 181.48. The first 10 cms of this locus were very similar to the contents of locus [LM23013]. It contained gravel, cobbles, and boulders, along with copious amounts of pottery. The pottery spanned from the 1st-5th century, but the majority of the sherds were tiny fragments of 1st/2nd century ESA ware, with only a few later sherds mixed in. Then on 23-6 at an elevation of 181.32 we encountered a layer basalt boulders across the whole probe. These boulders had been split, laid with the flat side down. The pottery obtained from beneath these stones was relatively clean compared to the pottery above and possessed a span from the 1st-3rd century. The latest datable material is 1 sherd of a Havarit cook pot, which spanned from the 1st to the 3rd century. On 26-6 at an elevation of 181.24 we hit another layer boulders throughout the entire probe. We removed most of the boulders except for the course of stones along the western balk. These stones were definitely laid intentionally in a line and may be part of a wall or wall foundation. Directly underneath these boulders at an elevation of 181.10 was a thin layer of crumbly dark grey cement. It is unlikely that this layer was a road since it was so thin and so crumbly. Therefore, this cement is probably the remnants of a building floor. The pottery retrieved from under the second course of boulders also dates from between the 1st- and 3rd centuries. It is probable that the wall of stones and the cement floor are part of a first century building. This suggests that the Romans moved the road after the first century, perhaps in the second century during the building of the second temple. This is an exciting new theory, but only further excavation in the area next season will confirm or refute this theory definitively.

This season was very successful in M23. We established an accurate, although preliminary, temporal phasing the road and structure in L23 spanning all the way from the 6th century to late 2nd/early 3rd. But the discovery of the possible 1st century building has posed a whole new set of questions that will need to be addressed next year.