Characterization of Brazilian agates containing a lower horizontally banded section and an upper chamber with bands parallel to the walls shows that these agates formed much later than the 135 Ma Paraná basalt host rock. Age differentiation between the two types of banding shows that the horizontal bands formed between 43 to 63 Ma ago with a final infill of wall-lining bands between 7 and 27 Ma later. The horizontal bands have a higher Al3+ concentration and a greater crystallite size than the wall-lining layers; they have a lower mogánite content and defect-site water content. The formation of these agates appears to be the result of a three-stage process. After the separate formation of horizontally banded and wall-lining agate, a silica infill seals the gap between the agate and the cavity wall. The detection of cristobalite in some specimens indicates that genesis of both the horizontally banded and wall-lining deposits in the Brazilian samples proceeds along an amorphous silica → opal-CT → opal-C → chalcedony pathway.

Dehydration of silanol and molecular water in 60 agates from 12 hosts with ages between 23 to 2717 Ma has been investigated using desiccators and high-temperature furnace heating. There are wide differences in the water data obtained under uncontrolled and fixed atmospheric water vapour pressure conditions. After agate acclimatization at 20°C and 46% relative humidity, the total water (silanol and molecular) was determined in powders and mini-cuboids by heating samples at 1200°C. Agates from hosts < 180 Ma all showed a greater mass loss using powders and demonstrate that after prolonged high-temperature heating, silanol water is partially-retained by the mini-cuboids. Desiccator dehydration of powders and slabs shows that powder preparation can produce water losses; this is particularly relevant in agates from hosts < 180 Ma. The identified problems have consequences for water quantification in agate and chalcedony using infrared or thermogravimetric techniques. Mobile and total water in agate is considered in relation to host-rock age, mogánite content and crystallite size. Links are observed between the various identified water contents allowing comment on quartz development and agate genesis. The water data also supports previous claims that agates from New Zealand and Brazil were formed long after their host.