296 THE BELL SYSTEM TECHNICAL JOURNAL, MARCH 1953 



be determined. Practical systems, as a matter of economy, are not 

 equipped with indicating mechanisms to show performance at all stages. 

 Events take place rapidly and the causes leading up to a particular traffic 

 situation cannot be easily observed. Traffic loads cannot be repeated 

 under controlled conditions. Size of installation and quantities of working 

 equipment can be varied only by small amounts in a working office. 

 To test offices of various sizes requires that these offices be built and 

 installed. Variations in the system operation require actual changes in 

 a working system. This can be done only to a limited extent. 



In the case of newly developed systems, estimates of performance can 

 be made on the basis of engineering judgment and experience with older 

 systems of a similar type. This can be followed by a trial installation of 

 a working system of a modest size which will test the system for flaws 

 in design as well as provide information on traffic capacity which can be 

 extrapolated to indicate approximate quantities of equipment for larger 

 installations. At best this is a slow and expensive process and engineering 

 data must be continually revised as experience is gained Avith larger 

 installations. When the system is to be used extensively in installations 

 of various sizes, methods of evaluating system performance in advance 

 of actual construction are desired. This situation occurred in the develop- 

 ment of the No. 5 crossbar system and was the occasion for building 

 the present throwdown machine. 



THROWDOWN TECHNIQUES 



Since throwdown techniques have played an important part in the 

 development of telephone traffic theory, a brief discussion of the basic 

 principles will be given. 



A single throwdown test will indicate the performance of a system 

 under a specific set of conditions. In a typical telephone traffic study, a 

 given traffic load would first be assumed and a simulated system installa- 

 tion to handle this load would be engineered on the basis of the best 

 available information. The test run then will show the performance of 

 the system under these particular conditions, and indicate both the 

 adequacy of the initial engineering procedures and possible improve- 

 ments. To obtain a proper balance between equipment quantities and 

 traffic load may require several additional runs, varying the equipment 

 quantities, traffic load or both. 



The procedure in a throwdown study i§ to first obtain data represent- 

 ing the traffic to be handled by the system. The traffic data can be 

 generated artificially by the use of random numbers. The method is 



