THE EARTHQUAKE PROBLEM HECK 365 



departures from uniform rotation are likely and, therefore, it is 

 desirable to have all components record on one drum or on several 

 drums on the same shaft (pi. 2, fig. 2), The same activitj'' vs^ill 

 not then be recorded at different times because of variation in speed 

 of several independent drums, and this is an advantage even if the 

 absolute time is in error. Time marks from a good clock are placed 

 on the record and time signals by radio are recorded directly on the 

 seismograms, and if this can be done with sufficient frequency errors 

 can be kept small (pi. 3, fig. 1). Eev. James B. Macelwane, S. J., 

 in a paper before the section of seismology. International Geodetic 

 and Geophysical Union, last summer stressed the importance of time 

 and showed that there is grave danger of assumption at many seis- 

 mological stations that the accuracy is greater than it actually is. 



Galvanometric recording systems. — In such systems the boom, in 

 addition to carrying the mass, has a coil which moves in the field 

 of a strong magnet. The currents set up are recorded through 

 a galvanometer. This system has the advantages that high mag- 

 nification can be readily obtained, damping is not difficult, and 

 tilting of the pier is not recorded on the record. Also, it is at times 

 a great advantage to have the seismometer at a place specially 

 suited to this purpose and the recorder at another place at some 

 distance away, where conditions are better suited for recording. 

 The distance may be considerable if conditions require it. 



In the case of most horizontal component instruments with long 

 periods, as 12-15 seconds or more, tilting of the pier may seriously 

 confuse the record. The effect is to put successive lines on the record 

 too far apart or, more serious, to crowd them so close that an earth- 

 quake record can not be interpreted. This tilt is usually due to 

 local temperature effects on the building or on the pier itself. The 

 effect of tilting is eliminated in the galvanometric recording system 

 and in the case of short period instruments, and it is very desirable 

 to eliminate it on other types. This has been accomplished success- 

 fully in an instrument which will be described later. 



Examples of instruments which emhody these pri?iciples. — The 

 Wood-Anderson torsion seismometer, through the use of a very small 

 cylindrical mass or a vane attached to a vertical fiber under tension 

 is able to get a short period with high magnification and so is able 

 to respond to and to magnify ground movements of very short 

 period. It is probably the best instrument in existence for the study 

 of near-by earthquakes. It was developed at Pasadena, Calif., for 

 the special studies that are being made there. Wliile ordinarily 

 operated at a period of less than 1 second, it has been operated at 

 6 seconds, and the Coast and Geodetic Survey is operating such an 

 instrument at Tucson, Ariz. It not only gives good records of dis- 



