IN NATURE, vol. xxiv. p. 113, there is a notice of a new seismometer which has several advantages claimed for it. Might I suggest what seems an obvious and important improvement? As a rule pendulums cannot record vertical or oblique motions, and yet these are often the most necessary and valuable to record. 1. To do this, and yet as easily allow of lateral registration, I would say, support a heavy (leaden) ball of some 100 lbs. by a 30 or 40 feet spiral or rubber spring of suitable strength. It will be found that a very considerable amount of vertical play can take place, especially vertical effort, ere the ball can be affected, and that lateral play of the support will produce very little effect indeed, unless, as is most unlikely, the motion is prolonged and is continuous in one direction. 2. Around the sphere, and at a very short distance from its surface, radial rods actuated like the key-plugs of a cornet are supported, say at every 30° all over the surface, contact with any one of which will electrically record time, and the pencil attached to the plunger record distance of stroke on revolving paper attached to plunger-tube.