Block D Temple
The above description is of the ruins themselves, but to create a three dimensional model we had to build up from the footprint. This was quite a challenge, as the only way to justify certain architetural features is through analogy to other sites.
We chose the height of the out enclosure wall rise up 5 meters from Main Street. Once inside the enclosure, then, the wall stands only 3.5 meters high as the whole area is on a 1.5 meter raised platform. The visitor enters by climbing up either side of the double stairway. Once inside, she can see the whole temple complex. A significant difference between Marshall’s discussion of the temple complex and our model pertains to the two “stupa bases” outside the main temple. Domenico Faccenna has made a very convincing argument that these types of bases were for pillars, not stupas, and we have chosen to follow Faccenna in this regard.  The pillars we constructed are 5 meters high and are topped with a circular disk.
Domenico Faccenna, “Columns at Dharmarajika (Taxila),” East and West 57, no. 1-4 (2007): pp. 127-173.
As for the temple itself, Marshall identified the whole structure as a Buddhist grhya stupa,  and Ahmad Dani agrees, rightly comparing it to contemporary rock cut grha stupas in “South India.”
Marshall, Taxila, pp. 152-154. Found in the temple were a number of fragmentary stucco sculptures, mostly Hellenized Buddha heads of which a significant portion seem to be female, but, unfortunately, not much else. Alexander Cunningham, in an earlier dig, found some interesting burnt clay figures as well. One was the fragment of a hand that measured 6.5" across the breadth of the figures, and the other was a head which had a face 10.5" in length. See also Alexander Cunningham, "Shahderi or Taxila," in Archaeological Survey of India: Report for the Year 1872-1872, ed. Alexander Cunningham (Calcutta: Office of the Superintendent of government Printing, 1875), p. 74, and Behrendt, The Buddhist Architecture of Gandhara, p. 71.
 This identification has not gone unchallenged, and most recently Luca Colliva has suggested that the complex might be dedicated to the deities of a Brahmanic or Saiva cult and contain sculptures of these deities on the inside along with images of the Kushana emperors on the outside of the temple.
Dani, The Historic City of Taxila, pp. 100-102. By “South India” we take Dani to mean the region of Maharashtra where the most prominent Buddhist rock-cut caitya halls are still extant.
 Certainly this is a possibility, but we are more convinced by Kurt Behrendt’s interpretation of the function and religious nature of the structure – an interpretation that sees the predominant usage as Buddhist, but not in the same way the Marshall and Dani imagine (see below). But before detailing these arguments, let us get back to the structure of the building. The visitor enters first a small, raised porch, and then enters the antechamber. The antechamber, then, was most likely covered with an arched roof made of wood, a free-standing version of the earlier Maharashtrian rock-cut cave caitya halls. Dani writes,
See Luca Colliva, “The Apsidal Temple of Taxila: Traditional Hypothesis and Possible New Interpretations,” in The Temple in South Asia, ed. Adam Hardy (London: British Association for South Asian Studies, The British Academy, 2007), pp. 21-27. Colliva (citing G. Verardi, “The Kushana Emperors as Cakravartins. Dynastic Art and Cults in India and Central Asia: History of a Theory, Clarifications and Refutations,” East and West 33, no. 1-4 (1983): pp. 225-282) compares this temple to the religious complex at Ma?, and suggests that, like the Ma? temple, images of SHri or Lakshmi as likely candidates for the central deities. Collliva (citing Pierfrancesco Callieri, “Buddhist Presence in the Urban Settlements of Swat, Second Century BCE to Fourth Century CE,” in Gandharan Buddhism: Archaeology, Art, Texts, ed. Pia Brancaccio and Kurt A Behrendt (Vancouver: UBC Press, 2006), p. 11) also suggests that the other stucco sculpturs might have been of Kushana emperors.
The recovery of large iron clamps, nails, and charcoal led Sir John Marshall to believe that the material of the roof was timber, and according to him this was the earliest example of its kind known to us, not only at Taxila, but in the whole Indus region. 
Dani, The Historic City of Taxila, p. 100-101. Such rock cut caves are well perserved at Maharashtran sites such as Karle.
This comparison to South Indian rock-cut temples led both Marshall and Dani to posit the existence of a stupa in the cella. However, three structural features of the temple argue against such an interpretation. First, unlike South Indian rock-cut cave grhya stupas, a large wall with a door separated the recangular antechamber from the back cella. That is, in the South Indian rock-cut caves, the cella was easily accessible, even visible, from the antechamber. Not so in the Block D temple. Second, the area running around the temple that Dani understands as a circumambulatory path, a staple feature of South India rock-cut grhya stupas, is just too narrow for circumambulation, and the circumambulatory path would be blocked by the separation wall anyways.  Third, and most importantly, if the cella did house a stupa, one would expect to find a stupa base, or some kind of evidence of it. Marshall argues that the lack of evidence is due to treasure hunters who opened the stupa looking for relics. But remember that the remains in the cella do not suggest some kind of haphazard digging, but rather contain what is a very deliberately dug pit—it is finished with plaster on its walls and a has stone floor. As Behrendt argues,
Behrendt, The Buddhist Architecture of Gandhara, p. 69.
I have examined many looted stupas (the reliquary has been a common goal of treasure seekers), and I have never found the the superstructure of the stupa had been leveled; moreover, looters would be most unlikely to dig a finished hole 9.75 m in diameter and 5.5 m deep. 
Ibid., p. 70 and n. 33.
Behrendt, while agreeing that the structure was Buddhist in nature, finds it improbable that the cella housed a stupa. Rather, he suggests that the structure functioned as a “direct-access relic shrine.” 
Kurt Behrendt has pioneered the understanding of these direct-access relic shrines, as he argues in Behrendt, The Buddhist Architecture of Gandhara, p. 61:
Until now, such shrines have gone unnoticed in the architectural record, but there is considerable evidence in texts, sculptural depictions, and architectural remains to provide us with insight into the function and importance of such structures. These 'direct-access' relic shrines were constructed at many Gandharan Buddhist sacred areas to provide place where important relics, like the alms bowl of the Buddha, could be seen, touched, and venerated. These shrines also had to be constructed in such a way that they provided security for the relics when they were not on display.
See also Kurt A. Behrendt, "Relic Shrines of Gandhara: A Reinterpretation of the Archaeological Evidence," in Gandharan Buddhism: Archaeology, Art, Texts, ed. Pia Brancaccio and Kurt A. Behrendt (Vancouver: UBC Press, 2006), pp. 83-103. It should be noted that not all scholars agree with Marshall, Dani, and Behrendt that the structure is Buddhist in nature. At the very same conference that Behrendt presented his interpretation of the Apsidal Temple, Pierfrancesco Callieri disagreed and suggested that the temple was perhaps a naga shrine. He points to both the apsidal shrine at Bhir Mound and the similarly constructed naga shrine at Mathura, see Callieri, "Buddhist Presence in the Urban Settlements of Swat," pp. 74-74 and notes 24-28. And as argued above, most recently, Luca Colliva has suggested that the structure had a Brahmanic/Shaiva function, see Colliva, “The Apsidal Temple,” pp. 21-27.
Behrendt argues that the cella was a storage area, cut off from the public and only accessible to ritual specialists, that housed, quite securely behind two massive doors, a number of important relics. 
For Behrendt’s full argument, see Behrendt, The Buddhist Architecture of Gandhara, pp. 68-72. Marshall suggests that the absense of the actual ruins of a stupa in the apse is due to the activity of looters. However, Behrendt doubts this as “I have never found
At the east end of the rectangular antechamber, then, there would be a “relic throne” on which certain relics would be temporarily placed for worship. Finally, we had to decide how the roof was supported. The rock-cut cave temples are not as helpful here, as the “roof” of these grhya stupas did not need support; these roofs were the rock of the cave itself. But there is an exampl of North Indian apsidal temple at Sonkh. Herbert Hartel dates this structure from its first phase in the first century BCE to its second phase in the early second century CE. 
H. Hartel, Excavations at Sonkh: 2500 Years of a Town in Mathura District (Berlin: Dietrich Reimer Verlag, 1993), pp. 413-436. See in particular the line drawings, photographs, and nice penciled reconstruction.
While this temple is significantly smaller than the Block D Temple, it too is on a raised platform. We, therefore, comibined the structural elements of the South Indian rock-cut caves with the apsidal temple at Sonkh. This seemed fairly straight forward. However, once the three dimensional model was created, we realized that there was no natural light coming into the Block D temple. Dani suggests that there might have been windows located in the “ambulatory path,” but as we have argued, there most likely was not ambulatory path. 
Dani, The Historic City of Taxila, p. 101.
Further, the walls are quite thick, more than 3.5 meters, and windows are unlikely. So, in our reconstruction, the columns which hold up the wooden roof stand 5 m high, but the outer wall of the temple stands at 4.7 m high. This allows for light to stream into the large antechamber, but also protects the relic throne and visitors from the rain. Like the Sonkh apsidal temple, we could add a extended border, perhaps tapered down, to further protect the inside of the temple.