Page 4. Pliny’s Glass Recipe (Plinius Secundus Maior, 23/24 – 79 n. Chr.)
Pliny reports a glass recipe as follows (Naturalis Historia 36, 189-204):
194 Haec fuit antiqua ratio vitri. iam vero et in Volturno amne Italiae harena alba nascens sex milium passuum litore inter Cumas atque Liternum, qua mollissima est, pila molave teritur. dein miscetur iii partibus nitri pondere vel mensura ac liquata in alias fornaces transfunditur. ibi fit massa quae vocatur hammonitrum atque haec recoquitur et fit vitrum purum ac massa vitri candidi. …
Translation (according to Loeb, Pliny VIII):
This was the old method of producing glass. Now, however, in Italy too a white sand which forms in the River Volturno is found along 6 miles of the seashore between Cuma and Literno. Whereever it is softest, it is taken to be ground in a mortar or mill. Then it is mixed with three parts of soda, either by weight or by measure, and after being fused is taken in its molten state to other furnaces. There it forms a lump known in Greek as “sand-soda”. This is again melted and forms pure glass, and is indeed a lump of clear colourless glass.
Until now all editors (with one exception, see below) interpreted these sentences spontaneously in such a way that three parts (= ¾) of soda have to be mixed with one part of sand (= ¼) to give 4/4 parts of mixture. The problem of this interpretation is that this mixture doesn’t lead to a solid glass but only to watersoluble water-glass. The recipe of Pliny therefore was regarded as being wrong, also because he did not mention the addition of lime which is necessary for producing an usuable glass.
The author of this site pointed out that according to many published analysis results Roman glass contains 65 – 73 % of silica. He analysed sand of the bank of River Volturno and found a calcium carbonate content of 24 % (a certain calcium content had to be expected because of the geological situation of the landscape). If you assume a sand addition of ¾ = 9/12 according to the old Roman weight system based on 12: 12 unciae = 1 libra, you will obtain a good glass with a silica content of approx.70 %, the accurate value depending on the sand actually used.
The interpretation that Pliny meant twelfths when saying parts was published already by W. Froehner (La verrerie antique (1879), 27), but his idea was not taken into account.
The Pliny Group editing Pliny’s Naturalis Historia accepted the author’s proposal to come back to Froehner’s interpretation. The Pliny edition 2000 of the group (R. C. A. Rottlaender (Ed.), Plinius Secundus d. Ae., Ueber Glas und Metalle. Uebersetzt und kommentiert von der Projektgruppe Plinius. St. Katharinen 2000) therefore speaks of Pliny’s parts as twelfths (= unciae).
In this way archaeochemistry solved an old philological problem.