There are three major sources of what can be called silent letters: sound changes, analogies drawn to those changes, and respellings intended to indicate a word’s etymology.
(1) Sound change: Most silent letters are due to changes in pronunciation over the centuries. Often the change in pronunciation leads to a concomitant change in spelling so that there is no silent letter. For instance, Old English myln became our mill after the [n] was lost, and “hand in cap” became handicap. But because the spoken language changes faster than the written language, changes in pronunciation are not always matched with changes in spelling, leading to what can be called silent letters. This is the case with words ending, for instance, with the complex spelling <mn> pronounced [m], such as condemn, hymn, column, solemn, autumn. In all of these words the <n> was originally pronounced, as it is now when we add a derivational suffix with an initial vowel, as in condemnation, hymnal, columnist, solemnity, autumnal. (But notice that the <n> remains silent when an inflectional suffix is added: condemned, condemning.)
One important class of sound changes resulted in the silent final <e>’s that we use as diacritics to mark long vowels, soft <c>’s and <g>’s, voiced <th>’s, and a few other cases. Most often these silent final <e>’s are the remnants of earlier inflectional endings that over the centuries reduced to a soft schwa sound spelled <e>, which in time fell silent. (For more on the sources and functions of silent final <e>, see American English Spelling, pp.145-160.)
And another important, though limited, sound change involves the <gh> that descends from two Old English velar fricatives similar to the sounds at the end of the Scottish pronunciation of loch and the German pronunciation of Bach. In Middle English these fricatives were spelled <gh>. But over time the fricatives either changed (as in laugh) or disappeared so the <gh>’s fell silent (as in sigh, night, ought). For more details look at the earlier question on this site concerning <gh>.
(2) Analogy: A similar case of sound change involves the <b> in words like >catacomb, climb, comb, dumb, lamb, plumb, tomb, womb, all due to lost [b] sounds. However, words like limb, thumb, crumb illustrate the second main source of silent letters: analogy. These words come from earlier words in which there was neither the sound [b] nor the letter <b>: As people saw the several and common words like bomb, they began to assume that there should be a final <b> on words that in Old English had been lim, thuma, and cruma, an instance of analogical thinking that leads to our limb, thumb, crumb, and a few others.
(3) Etymological respellings: A third source of silent letters is the respelling of words to bring them more into line with their Latin and Greek sources, either true or assumed. Thus, Middle English det, dette was respelled in the 16th century to debt to reflect its Latin source, which is also the source of our word debit. (See the earlier question on debt.) There were similar respellings of doubt, redoubt, and subtle, which earlier had been douten, redoute, soutil. Similar etymologically-driven respellings account for the complex spellings of [t] in indict, receipt, and, erroneously, ptarmigan. It is doubtful that the silent letters in these complex spellings ever were pronounced.
These three sources can account for probably all of the so-called silent letters in English: the <w> in two
, the <l> in salmon
, the <p> in cupboard
, the <b> in subpoena
, the <l> in should
, the <t> in often