In 1937 Alec Reeves of Britain invented modern digital transmission when he developed Pulse Code Modulation. I say modern because Morse code and its variants are also digital: organized on and off pulses of electrical energy that convey information. While PCM took decades to implement, the advent of digital working was a momentous event and deserves much consideration. David Robertson, a biographer of Reeves, goes so far as to claim Reeves as the father of modern telecommunications. "I think a fair argument can be sustained that the adoption of digital is the principal motor of change in the early 21st century. For sure, there'd have been no merger between AOL and Time Warner and other moves towards combining media with telecom companies had it not become possible to transmit information of all sorts in the same binary way. Whether all this is good news is, of course, another issue."
                 Source
http://www.privateline.com/TelephoneHistory/History1.htm

1     First telephone numbers are just names
2     Depending on exchange size, two, three or four digit numbers assigned to subscribers, 
3     Two letter prefix codes assigned to four digit numbers (Circa 1928 to 1958) 
4     In larger cities three letter prefix codes assigned to four digit numbers (Post WWII) 
5     Seven digit, all number dialing begins phase in. (1958) 
6     Nearly all of North American telephone network converted to all number dialing (1985) 
7     Some party lines remain, with single digits like Rodeo Creek Number 8

People in manual exchanges placed their calls first through an operator, who listened to the number the caller wanted and then connected the parties together. 

The first telephone numbers weren't numbers, they were names. The name of your company or you as an individual. That was too confusing to build a telephone system on since many people in a town might share the same name. Starting in 1879, then, scarcely three years after the telephone was invented, the switch to assigning a customer a number began, with a four digit code being typical. Calls were not dialed by the customer, indeed, there were no dial telephones yet. All calls were connected manually by an operator at a switchboard. But dial telephones would come along.

The Bell System thought dial tone a good substitute for an operator's "Number please" and required this service in all of their automatic exchanges. Before the 1950s most of the independent telephone companies, but not all, also provided dial tone. And, of course, dial tone was not possible on phones such as crank models, in which you signaled an operator who then later connected your call.

AT&T's operating companies started installing dial telephones in the mid to late 1920s. Customers could now dial numbers themselves, instead of having an operator place them as before. Rather than use all digits to indicate a telephone number, AT&T hit upon a hybrid system of letters and numbers. Instead of a number like 351-1017, the Bell System referred to it by a name like ELgin 1-1017, ELliot 1-1017, or ELmwood 1-1017. Something like that. The two letters and a number indicated a customer's switching office or exchange, the last four digits the actual customer's number. But why use letters?

I've never understood, though, why PEnsylvania 6-5000 should be easier to remember or dial than 436-5000. Yet for forty years the most bizarre exchange names flooded the country and the entire telephone system was
based on this riot of numbers and letters. It's with some satisfaction I note that AT&T's Joel and Schindler, in A History of Engineering and Science in the Bell System: Switching Technology (1925 -- 1975), in discussing the Texas trial above, state contritely that "later human-factors studies showed there was no need for letters in the dialing sequence." Whoops! They went on to say that people in 1958 were now used to dialing, quite unlike forty years before. Four decades of practice were needed before people could dial another two or three digits? Perhaps. But I doubt it.

            Source
http://ourwebhome.com/TENP/Recommended.html