These fragments are usually curved and bear the script upon the convex side, care should therefore be taken that they are completely immersed, and that no large air-bubbles prevent the access of the water to any part of the under-surface. The writing is done with either lamp-black or more rarely some form of iron ink, and is retained mechanically by the porous character of the ostraca. In the latter case the characters may be enhanced by the application of a dilute solution of tannic acid, which sometimes proves useful also for limestone pieces.

If these fragments are sufficiently few in number to allow each to be put into a separate glass vessel, the washing out of the salts is completed so quickly that there need be little danger of obscuring the script. When large numbers were to be washed and when the script was already indistinct I have employed the following method: After examination as to their fitness for immersion the fragments are placed on a wooden grating in a tub, in which they remain for a couple of days, during which the water is renewed once. They are then taken out and allowed to dry. All those which still show the script distinctly are separated and their steeping is completed, but the remainder, having been completely dried, perhaps on the top of a warm stove, are brushed over once or twice with a dilute (1:6) mixture of varnish and benzine in such a way that the surface is only moistened, and when dry shows no gloss. The pieces thus superficially varnished are kept in a dry place for about two months, until the varnish is hardened; the process of washing out the salt is then begun again. The thin coat of varnish fixes the script without interfering with the steeping. The varnish solution must be dilute, for a thick coating will partially peel off from the object in the course of the steeping, or will remain in the pores in the form of opaque particles, and thus render the script illegible.

The same difficulty which arose in the treatment of the Meten limestones was frequently met with in the treatment of these ostraca. Those which were of a dark brown colour especially, and to a less degree also the red and the yellow, were covered with a slimy growth of algae. As the script is easily destroyed no attempt should be made to remove these algae from the side which bears the script even with the softest brush, although they should from time to time during steeping be brushed from the underside. The inconvenience caused by algae is, however, less marked in the treatment of earthenware, the light and porous character of which renders prolonged steeping needless, nor is there the same necessity to continue the steeping for the purpose of chlorine estimation. The following results were obtained in the treatment of 13 fragments, the average thickness of which was 1 cm. [3⁄8th inch], with an average superficial area of 1⁄10 sq. metre [4 inches]. The tub in which they were steeped contained 85 litres [181⁄2 gallons] of water.

100 cubic centimetres of the tap-water used were found to require 0·5 c.c. of the silver solution, and on each occasion this quantity of the water was tested.

Water changed after111122245 days
100 c.c. of the water used in steeping required3·01·31·00·80·90·70·70·60·6 c.c. of silver solution

The water was changed at first daily, then every two days, and so on: the steeping could therefore be regarded as complete at the end of a fortnight.

A small figure of earthenware, which weighed only 28·9 grammes, was steeped in 11⁄2 litres of distilled water, and gave the following result for every 100 c.c. used:

Water changed after 2 days required 3·6 c.c. silver solution.
Water changed after 3 days required 0·4 c.c. silver solution.
Water changed after 4 days required 0·0 c.c. silver solution.

The steeping was, therefore, in reality complete after five days, and, as the steeping water was thoroughly mixed before the withdrawal of the 100 c.c., the total quantity of sodium chloride contained in the figure can be calculated as follows:

For the 15th part (viz. 100 c.c.) of the water 3·6 + 0·4, i.e. 4·0 c.c., of decinormal silver solution were used, which is equivalent to 15 × 4·0, i.e. 60 c.c., of silver solution for the whole quantity. Now 1 c.c. of this decinormal solution corresponds to 0·00584 gramme of sodium chloride; the water therefore contained 60 × 0·00584 gr., or 0·35 gr. sodium chloride. Thus the figure contained altogether 11⁄5% of sodium chloride.