On p.557 of Steven Landburg’s Price Theory and Applications (6th international edition), I find this:
It is widely believed that Treasury bills carry essentially no default risk and that the U.S. Treasury has never defaulted on its obligations. This is untrue. For example, the Treasury defaulted on bill #GS7-2-179-46-6606-1.
In order to purchase a Treasury bill at auction, the investor (that is, the buyer of the bond) must submit a payment equal to the full face value of the bond. Following the auction, the discount is supposed to be returned to the investor immediately. [...]
One unfortunate investor followed this procedure. His discount, approximately $1,100, was not returned. Following a series of inquiries, the Treasury took the remarkable position that although the default was entirely due to its own clerical errors, there was a strong possibility that the errors were irreparable and that the discount would never be paid. It required nearly nine months, considerable expense on the investor’s part, and the intervention of several senators and congressmen before the Treasury met its obligation. Even then, the Treasury refused to pay interest for the nine months in which it unlawfully held the funds.
The frequency of such occurences is not known. This particular investor went on to write a textbook on price theory, yielding a bit more publicity than might ordinarily be expected. [...]
Only a few pages later (p. 562), Landsburg claims that we are to be indifferent between taxation and government debt as means to government spending (not considering the object of spending) partially because “You can earn the same rate on your savings by the simple expedient of buying Treasury bills.” The cognitive dissonance is deafening.