How did the word come into English with the two variants czar and tsar? The 'ts' spelling is a transliteration of the Russian 'царь', but the 'cz' spelling is what interests me more. To me it looks Polish, where 'cz' is common, but is pronounced as English 'ch'. Where did this second form come from?
Here’s what the OED says:
The Slavonic word ultimately represents Latin Cæsar, but came . . . through the medium of a Germanic language in which the word had the general sense ‘emperor’ . . . The spelling with cz- is against the usage of all Slavonic languages; the word was so spelt by . . . the chief early source of knowledge as to Russia in Western Europe, whence it passed into the Western Languages generally; in some of these it is now old-fashioned; the usual German form is now zar ; French adopted tsar during the 19th cent. This also became frequent in English towards the end of that century, having been adopted by the Times newspaper as the most suitable English spelling.
Origin of 'czar' and 'tsar'
Merriam-Webster's Eleventh Collegiate Dictionary (2003) lists czar as the primary spelling in U.S. English, and tsar as a variant spelling. Webster's Word Histories (1989), in a fairly detailed discussion of the two spellings, attributes the spelling czar to a sixteenth-century Austrian baron:
czar Czar, or tsar, is our English word for a pre-Soviet Russian emperor. Tsar is a straightforward borrowing from the Russian, but the form of czar is strange. It looks rather like a Polish word, and in fact there is a Polish czar, but it is pronounced like English char and means 'charm' or 'spell'. The Polish equivalent of Russian tsar is spelled car—Polish c is pronounced ts. We owe our peculiar spelling of czar to an Austrian diplomatist, Siegmund, Freiherr (Baron) von Herberstein (1486–1566). ... Herberstein wrote in Latin, but his spelling of Russian tsar was influenced by his native German. The c in Herberstein's czar may have come from Polish, but his z was surely added as a pronunciation indicator—z in German, like c in Polish, is pronounced ts. The English word czar first appeared in a 1555 translation of Herberstein's work [Rerum moscovitarum commentarii, or Commentaries on Muscovite Matters].
The entry for the term in John Ayto, Arcade Dictionary of Word Origins (1990) focuses exclusively on the derivation of the spelling tsar:
tsar Caesar was a Roman cognomen (English gets caesarian from it) and from the days of Augustus was used as part of the title of 'emperor.' The Germanic peoples took it over in this sense (it is the source of German kaiser) and passed it on to prehistoric Slavic as tsēsari. This has evolved into Serbo-Croat and Bulgarian tsar and Russian tsar'—source of English tsar.
British and U.S. preferences for 'tsar' versus 'czar'
The question of whether tsar or czar should be viewed as the primary spelling in English is a matter of contention, as various answers and comments here indicate. Robert Hendrickson, The Facts on File Encyclopedia of Word and Phrase Origins (1997) offers this remark at the end of its entry for czar; tsar; tsarina; etc.:
Spelling of the word varies in both British and American dictionaries—neither form is accepted by all authorities in either country.
That the tendency of reference works to prefer tsar in UK English and czar in U.S. English is clear from the way the British and American versions of the OED handle the spelling. From Concise Oxford English Dictionary, revised tenth edition (2001)):
czar etc. n. variant spelling of TSAR etc.
From Oxford American Dictionary and Thesaurus (2003):
tsar var. of CZAR.
Bryan Garner, Garner's Modern American Usage, second edition (2003) comes out squarely in favor of the spelling czar:
czar; tsar. The spelling czar is overwhelmingly predominant. Tsar, though closer to the Russian form, is archaic.
And so does The Associated Press Stylebook and Briefing on Media Law (2002):
tsar Use czar.
The spelling czar no doubt also benefits in the United States from being identified as a probable typo under Microsoft Word's default U.S. spelling preferences.
And yet something very odd occurs when you run Google Books searches for tsar (red line) versus czar (blue line). The British English chart for the years 1809–2008 is not surprising, given the U.K. preference for tsar:
But the corresponding chart for American English is astonishing:
Despite being identified as the primary spelling by the Merriam-Webster, Oxford American, American Heritage, and Random House English dictionaries, tsar appears to have held its own against czar for the past century in the Google Books database of U.S. publications.
A usage note under the entry for czar in The American Heritage Dictionary of the English Language, fourth edition (2000) offers perhaps the best explanation of this seeming anomaly:
The word czar can also be spelled tsar. Czar is the most common form in American usage and the one nearly always employed in the extended senses "any tyrant" or informally "one in authority." But tsar is preferred by most scholars of Slavic studies as a more accurate transliteration of the Russian and is often found in scholarly writing with reference to one of the Russian emperors.
To the extent that tsar continues to be at least as common as czar in U.S. books, AHDEL suggests, this may reflect the strong preference of Slavic studies authors for tsar. Presumably, if the Google Books database were to add newspapers and (more) magazines to its scanned holdings, the influence of Slavic studies scholars' preferences on the reported overall popularity of tsar versus czar in the United States would be much diminished.
Czar is not the only older form...
In 17th century texts both Czar and Tzar occurs. In this account of diplomacy from 1669 - A Relation of Three Embassies From his Sacred Majestie Charles II - : both forms occur, 'the czar' 5 times and 'the tzar' 56 times
But amongst all the Princes of Europe, that, by their congratulations of his Re-establishment, seemed ardently to aspire at His Alliance, the Tzar of Moscovy had the most equitable pretentions. ― p 2
..by Moscovie are ordinarily understood all the Provinces united under the Obedience of the Czar.
― p 26
'the Czar' occurs in 5 places between pages 26 and 47 while 'the Tzar' is the only form in the rest of the book, as well in the index. This might reflect the usage of two different writers..
'Tzar' is also used in a Swedish text from the same period: Joh. Widekindi. (1671) Thet Swenska i Ryssland Tijo åhrs Krijgz-Historie:
och then Store Herre Tzar och Storfurste Michael Fedorowitz ― p 883 ...
As late as in James Bell (1850) A System of Geography, Popular and Scientific, both Tzar and Czar occurs (5 times each)
and in New Monthly Magazine volume 78 (1846)
'the czar' twice (244, 245)
'the tzar' five times (p 26, 38, 31; 75, 90)
'the tsar' twice (31, 38) [A Summer in Russia Chap VI]
It seems the spelling 'czar' was first used by Sigmund Freiherr von Herberstein who was twice envoy of the Holy Roman Emperor to Russia. In 1549 life he published his notes about the lands he visited in Rerum Moscoviticarum Commentarii.
Czar Rhutenica lingua regem significat. cum autem communi Slavonica lingua, apud Polonos, Bohemos, & alios omnes sumpta quadam consonantia, ab ultima, & ea gravi quidem syllaba Czar, Imperator seu Caesar intelligatur -p18
von Herberstein spoke Slovenian, so it might be fair to assume that spelling he used would be a Slovenian approximation of word he heard (or read).
This is pure speculation on my part but maybe what he heard was some Russian pronounciation of OCS цѣсарь, perhaps [tsəsar]
The spelling 'czar' was almost universally used in the U.S. well into my adulthood. I am now in my 80's. During WWII we had various people in the government running imnportant programs who were dubbed 'czar' of their program, whether it had to do with manpower or some industry or price controls. The word morphed into 'tsar', which my limited knowledge of Slavic languages would suggest is closer to the Russian, sometime later, probably gradually through the '50's to 70's--much more recently than a century ago. I can recall noting the change and wondering why. If the OED doesn't cite usage to indicate when the change took place, I don't know who would.
I live in America, where the "Czar" spelling was commonly used into my adulthood - in the late 1980's! As noted already by Ray and Fumbefingers, OED cannot be accurate or apply accurately to American English - "the end of the (19th) century" was definitely not in "the past 50 years". The Polish in "Czar" may have contributed to the "Cz" usage in the US, where a large Polish-American population had existed since the 19th century. That it's been changing here during the few decades since the Soviet collapse sent a new population of Russian immigrants our way is hardly surprising.