The teeth of the key of the Roman lock described above, it will be seen, are made to fit exactly the holes in the bolt; and this may perhaps have served to give the first idea of the ward system, which was so greatly depended upon for security in later times; but the same fallacy attaches to the use of these fitting teeth which attached to the ward system generally, for it is evident that any form of tooth small enough to go into the holes, and of the proper length, would have sufficed to lift the tumblers and draw the bolt; and accordingly we find that, in the Roman key usually discovered, the teeth are merely round pins, and have no particular form given to them for fitting purposes. The distribution of this class of lock may be determined by the localities in which the keys and bolts have been found. Fig. 22B, Plate III., is a bronze bolt of this description in my collection, from Oppenheim, and obtained by me at Mayence. Fig. 23B, Plate III., is another of bronze, also in my collection, from Heddernheim. Similar ones have been found repeatedly in France, Italy, Germany, Switzerland, and England. The keys with teeth are even more widely distributed, and have been found in all those countries which have been occupied by the Romans. Fig. 24B, Plate III., is a large iron key of this description in my collection, found in the Rhine, at Mayence. The earliest known example of a key with teeth, according to M. LIGER, is one represented on a coin of the PAPIA family, dating about the end of the 2nd century B.C.[7] [7] 'La Ferronnerie,' tome i., p. 261, fig. 208. But the ward system appears to have developed itself still further in connection with these locks and before the revolving key was introduced. Fig. 25B, Plate III., is a specimen of a class of keys frequently discovered with Roman remains, in which a plate is attached at right angles to the pins. This plate is pierced with slits of various forms, apparently intended to admit of the passage of wards placed vertically beneath the bolt to prevent any but the proper key from rising to lift the tumblers. The direction in which these keys were raised is shown by the flat part of the handle of the key being always at right angles to the pins and in the same plane as the ward plate. Besides the bolts with several tumbler holes in them, others adapted for single tumblers have been discovered. Of these fig. 26B, Plate III., drawn from M. LIGER'S work, and found in the forest of Compiègne, is an example, and fig. 27B, Plate III., from the same work, and found at Nonfous, in Switzerland (Bonstetten) is a key adapted to fit such a bolt.