this stroke will give an approximation to the strength of the shocks. The middle of the stroke, or, if they are vertical, the end of the uninterrupted line, gives the time of the commencement of the shock. The strength and direction of the shocks may also be approximated if the (as respects rotation) fixed plate hihl have an annular recess, filled with quicksilver until its surface reaches the holes sss, made in the cylindrical sides. At the first motion of the pendulum, the quicksilver will be shed out through these holes into a dish divided into the same number of compartments as there are holes, like those already in use in many existing instruments of this kind (Cacciatores)".

Such are the chief seismometers hitherto proposed. They all involve in some form the principle either of the solid or of the fluid pendulum, the latter term being applied to the oscillations of liquids in tubes or other such vessels; and have disadvantages, both theoretic and practical or constructive, which render their indications inaccurate.

Every pendulum seismometer has a time of oscillation due to its length, which in the case of the solid pendulum is

I being the length of the pendulum and of the oscillating column of liquid respectively; but if P = the period of the earth-wave or shock, then whenever T=P, or n x P, or -, P, the indication of the instrument will be in excess

n

of the horizontal component of the wave's motion; when, on the contrary, T represents no function of P, it may be much less than it.

The amount of error depends also upon the velocity of movement of the horizontal component of the wave. If this be considerable, the solid pendulum, whether hanging or inverted, acted on by gravity or elasticity, is at the first moment left behind; as the rod becomes more oblique, the pendulum is dragged along, and acquires a velocity (in a direction which approaches to horizontal) greater than that due to the arc through which the pendulum has fallen in the time. At the end of the wave's forward movement, then, the pendulum is thrown forward too far; and at the end of the return movement of the wave, it moves beyond the range of the latter, by a small arc due to its proper motion. This objection applies, though with less cogency, to the fluid pendula, and in their case to both the vertical and horizontal components of the wave.

These discrepancies of indication will vary whenever the velocity and dimensions of the earth-wave become altered; and as, for the same instrument, T varies with sin2 A (A being the latitude), it is obvious that even two perfectly similar instruments at stations north and south of each other, will not give strictly comparable results for the same earth-wave.

These are but examples of one or two points of theoretic difficulty, to which others might be added, and which affect these instruments principally as indicators of the dimensions of the earth-wave. Some of these theoretic disturbances may be eliminated by calculation from the results; but there are also some apparently insuperable difficulties, of a practical or constructive nature, which affect all solid pendula as reliable indicators even of the direction of surface-transit (horizontal component) of the earth-wave. However finely suspended the pendulum-if acted on by gravity only, or, however constructed if by elasticity or by elasticity and gravity, it is found impracticable to produce an instrument that shall make even the second half of its very first complete vibration strictly in the plane of the original disturbance, i. e. in that of the wave's transit. If, for example, any one of the

Fig. 4.

d

e

instruments 5 (b), 6 (b), or 7 (b), be caused to make a semivibration by a movement of the nature of one horizontal jerk, and strictly in one vertical plane ab (fig. 4), the trace made will in most instances be found thus; cd, the first semivibration, is made sensibly in the plane of movement, but the returning complete vibration de, is found diverging from it through a sensible angleede. If the vibration of the instrument be suffered to continue, its trace rapidly becomes an extremely elongated ellipse, whose excentricity constantly diminishes, as well as the actual dimensions of both its axes, until the instrument comes to rest, after tracing thus a mass of elliptic spirals, from which nothing certain can be gathered as to direction in some instancesin which, at best, it is only possible to arrive at a probable direction of originating impulse, by drawing a mean major axis through all these closed

curves.

Constructively, this evil arises not only from the nature of the suspension, if a pendulum of gravity, or, if one of elasticity, from the form, material, &c. of the suspending or supporting spring; but also, in both sorts, from the fact that it is practically impossible that the point of suspension (or, in the spring, its centre of resistance), the centre of oscillation, and the resultant of the various opposing forces of the stile or tracing-point, shall lie in one vertical plane, and that that plane shall always coincide with that of the wave's movement; and hence lateral divergence of the pendulum and elliptic spiral oscillation. But it is also partly due to the nature of the earth-wave motion itself, which is never a purely normal one, but always more or less disturbed by small transversals; so that the initial movement impressed upon the pendulum is really not exactly that of the wave's transit. Before entering further, however, upon the subject of the actual perturbations of the superficial earth-wave, as now known, and their effects in relation to seismometers, some remarks may be advisable as to the special objections which I have either observed or experimentally ascertained in respect to each particular arrangement of the seismometers already described.

1 (a). The Cacciatore mercurial dish. If the earth-wave emerge with a considerable angle from the horizon, and large velocity, the mercury first surges up at the side of the dish towards which the earth-wave is in transit, and in the direction opposite to its motion; it then, after spilling out some of the mercury, commences its return oscillation, moving in the same direction as the earth-wave, and spills out another portion at the opposite side of the dish. The sum of the weights so spilled out, taken at either side of a diameter transverse to the earth-wave's vertical plane of transit, will vary with every change in the angle of emergence, or in the velocity or in the dimensions of the earth-wave. Small transversal vibrations, arriving almost along with the earth-wave, as well as the effects of the form of the dish, and of its delivering-spouts or adjutages, disturb the initial simple surge of the mercury across the diameter of the dish, and produce reflected and other secondary surge movements of the mercury, which traverse round the circumference of the dish, and spill out more mercury in irregular gulps. The final result is, that no reliance whatever can be placed upon its final indication, as to the plane of the earth-wave transit having passed through the centre of gravity of that semicircle of cups which are found to contain the most mercury. The result is not materially different if the line of transit of the earthwave be perfectly horizontal. This instrument gives no information whatever beyond a most uncertain approximation to the direction of the horizontal component of the earth-wave transit.

2(a). The same objections generally apply to this form of instrument, and one in addition, viz. that a viscid liquid like molasses must always give indications short of the truth as to excursion in the dish due to any given shock, and the more so as it is more tenacious and approaches nearer to a solid; and as we have no correct means of measuring viscidity, even assuming it constant for the same liquid, nor any certainty that the specific gravity of such liquids remains constant (it is certain molasses will not remain of the same density in any climate for any considerable length of time), so observations made through their means at different times and places can never be comparable.

3 (a). The same objections that apply to 1 (a) apply to the tub of coloured water, but in a mitigated degree, the diameter being large, the volume and depth of the liquid great, and the cylindrical sides of the tub free from any apertures or inequalities. The initial surge gives a much more distinct indication of direction than in either of the preceding instruments; and it does not very frequently happen that a diameter may not be found approximating, with tolerable certainty, to the plane of earth-wave transit. But in cases where the normal wave is preceded or accompanied by very appreciable transversals, those violent tremors that are now known as the frequent accompaniments of the actual shock-the water-tub seismometer will give no indication, or an uncertain one, unless watched and remarked as to transit-direction at the instant of the occurrence of the shock.

4 (a). Tubes partially filled with mercury give almost unobjectionable indications as to direction of transit. Their evils are too great delicacy or sensitiveness, for the observation of that class of earthquakes of mean power, which are the most important to be studied, and by which they are completely deranged occasionally, while they are continually being disturbed in such a seismic region by small tremulous movements that are unimportant to notice. As respects their indications of velocity and dimensions of the wave, they are liable to the objections already noticed as applicable to all pendula.

5 (6) and 6 (6). The main disadvantages of these constructions, viz. the suspended and the inverted solid pendulum have been already pointed out; it may be added here, however, that with the inverted pendulum of Forbes, the supporting spring is more or less crippled down, by a sharp vertically (or nearly vertically) emergent shock, which gives a lateral movement (greater or less) to the pendulum, as though

from a horizontal originating motion, so that the instrument gives in such cases an absolutely false indication.

7 (6). Mr. Budge's inverted spring pendulum, restrained to a single semioscillation in one plane, offers some decisive advantages over any other form hitherto proposed of the pendulum seismometer. The whole length of the pendulum is elastic; and the rod being light, the whole weight by whose inertia it is bent may be considered as in the ball or bob. If be the moment of resilience of the rod, and the deflection be not very great, the angle wpn=0, then

L being the length, and b the horizontal ordinate of deflection of the pendulum. It is plain that although, like every other elastic rod, this will have a time of vibration of its own, and be therefore liable to part of the theoretic objections made to the simple pendulum on the same account, this form of pendulum will be "brought up" much more nearly within the true limits of the earth-wave amplitude in its horizontal component.

Perhaps the ratchet and pall may not be the best mode, practically, of arresting its movement at the end of its first semioscillation, with sufficient delicacy, and other methods are obvious that may be applicable; but if the elastic rod be a flat plate of sufficient breadth in relation to its thickness, and each rod or pendulum (of the four) be so placed, with reference to the cardinal points, that its broadest dimension shall be transverse to its normal plane of flexure, it is then obvious that practically we may neglect any flexion of the rod edgeways, the four rods in section being posited thus (fig. 5)

and that thus we obtain a flexure, for each pendulum, practically limited to its own vertical plane of oscillation, and so can obtain, for any intermediate line of wave-transit between the cardinal points a good approximate resultant direction from the two adjacent component deflexions. Perhaps a flat ribbon-like rod of tempered steel, whose section should be a rectangle, with sides having the proportion of about 30:1, would be better than an elastic wooden lath; and in either case, it is probable that a tape or silk ribbon, fastened at the side r, and passing with friction through a small horizontal slot in the elastic rod, so as to be stretched by its deflexion and pulled through, would be the best and simplest mode of registering the deflexion, or the angle 0.

While this appears to me the best of the solid-pendulum arrangements, I do not wish to be understood as recommending any one of the class.

8 (b). Santi's arrangement is of course subject to the objections made to all pendula. It possesses some advantage in separation of the results in different azimuths, and therein in clearness of indication; but it also has special disadvantages of its own. If, for example, the line of earth-wave transit be from S. to N., and the E. and W. pendulum be set up at the S. side of its own wall, it will tend to be thrown off or out from the wall by the shock; if placed on the N. side of its own wall, its friction will be increased on its suspensions, and tracing-point, by its being thrown in or pressed against the wall; and if the line of earth-wave transit be, say N.W. and S.E., both pendula will be either thrown out from or pressed in against their respective walls, according to which side of the N. and S. walls they be fixed at. This source of variable inaccuracy might perhaps be eliminated by a double set of pendula, viz. one at each of the opposite sides of the N. and S. and of the E. and W. walls, which would thus be oppositely affected (in excess and in defect) by this source of error.

9(b). What has been already stated, with reference to errors common to all pendula, and the remarks made under 7 (6) as to the superiority of elastic over simple pendula, render it needless to enlarge on those which were only proposed as extemporaneous instruments, and for which they will be found convenient and useful, and not more inaccurate than much more elaborate ones.

Referring now to the second class, or self-regulating instruments, the disadvantage of the one

2(a), proposed by the author is of the same character as that of 4 (a) of the first class, viz. too delicate a sensitiveness to small tremulous shocks, which derange the composure of the instrument, without its giving decisive indications. The galvanic recording part of the apparatus was all that could be desired, and is of course applicable to other forms of instrument as respects the displacement portions. Indeed, apparatus identical in all its main characteristics has been since brought into successful and constant use by Professor Airy, Astronomer Royal, for the registration of astronomical and other kindred observations, and also by several experimenters abroad. An account of many such arrangements will be found in De la Rive's 'Treatise on Electricity.'

2(a). The same remark, I think, may apply to Professor Palmieri's seismometer, with this addition: the movement of the mercury, equal columns of which are contained in the opposite legs of each U-shaped tube, depends in his instrument wholly upon the U-tube being canted over more or less in its own plane, so as to throw the legs of the tube out of plumb. This, Professor Palmieri (if I do not misunderstand him) considers an inevitable consequence of the transit of the earth-wave at the instrument, conceiving the earth's surface to suffer, in every case, such a sensible heaving undulation, as to rock the instrument upon it, like a ship upon a heavy ground-swell. I must confess to entertaining great doubts that, in the great majority of earthquakes, any such sensible undulation (enough, at least, to produce a consible throwing out of plumb of the U-tubes) can occur, althoug I have no reason to doubt that, from its delicate sensitiveness, contact will be broken, and the instrument act in so far, by some of the violent jars or jerks that it may receive. This peculiarity constitutes, in fact, the essential difference in arrangement between the author's seismometer and Prof. Palmieri's. In the former the

1858.

G