Many of the comments evoke different "orphan works" scenarios. It is important to distinguish three categories:
1. Owner cannot be identified.
2. Owner identified, but does not respond to requests.
3. Owner identified and responds, but refuses permission, or imposes conditions with which the requester cannot or does not wish to comply.
Works coming within the first category can properly be labeled "orphan," so long as the endeavors to locate the owner are sufficient (see below). The third category does not, however, describe an "orphan work." Copyright includes the right to determine whether or not, and how, to authorize a given exploitation, as well as to select the exploiter. Both Berne and TRIPs posit that the author's rights are exclusive; exceptions and limitations such as compulsory licenses, are narrowly constrained and do not call into question the author's basic right of control.
The second category is the most problematic, because a copyright owner's right to refuse permission can also entitle it to be discourteous. But an unresponsive right holder may not be a right holder at all. The "owner" may be failing to respond because it is not in fact the pertinent right holder, or has died or gone out of business, and the request was not transmitted to the successor-in-title. Nonresponsiveness might best be analyzed as probative of inability to locate the right holder, rather than as a criterion independently justifying the labeling of a work as "orphaned." The latter course risks coming into conflict with the right holder's prerogatives.1
In the spirit of that reply, which looks not just to domestic statutory law but to treaty obligations and other nations' interpretations, what happens when a musicologist turns a clearly public-domain work into a playable form, a record company commissions recordings from that form, and the musicologist wants to be paid as the copyright holder? This should be an obvious result: Just as with any other translation, the translator has copyright in his/her version of the work. If a publisher doesn't want to pay the estate of I.A. Richards for use of his translation of Plato's Republic, it can reprint one of three easily available public domain versions or arrange for a different translator.
But isn't music different from translation from a foreign language? Perhaps in form, but probably not in substance. As you might have guessed by now, this is not a hypothetical instance. The Law Lords got it right in Hyperion Records, Ltd. v. Sawkins,  EWCA Civ 565 (19 May 2005), holding that at least on the facts of the particular case, Dr Sawkins was a translator, not a mere transcriber. Unfortunately, the court (particularly Lord Justice Mummery at ¶ 16) persists in using language parallel to the "sweat of the brow" in describing what Dr Sawkins provided: "effort, time, and skill." Lord Justice Jacob, while concurring with Lord Justice Mummery, goes a long way toward correcting the damage at ¶¶ 7782 (ironically enough, citing Professor Ginsburg!), and concludes:
86. I therefore think Patten J was right and agree that this appeal should be dismissed. I would add two points:
i) This sort of question (sufficient work to be "original") is just the sort of value judgment by a trial judge with which the Court of Appeal should be slow to interfere, see Biogen v Medeva  RPC 1 at ¶ 45. Some error of principle is called for [to reverse]. Here there was none.
ii) The solution accords with a reasonable view of public policy that the sort of work done by Dr Sawkins should be encouraged. It saves others the time and trouble of re-creation of near-lost works, but in no sense creates monopoly in them. If someone wants to use Dr Sawkins' short cut, they need his permission.
Very nicely, very elegantly, and very clearly stated. Note, too, that it covers most of the "more than copying" uses to which some purportedly orphaned works might be putconsider a Ken Burns-style history of the civil rights movement to be illustrated using old family photographs whose origin is obscure. But that argument is for another time.