Their convictions after proceedings perceived to be fair would vindicate the values of clean government, deter potential wrongdoers and heighten confidence in courts that began to free themselves from decades of authoritarian Kuomintang government fewer than 20 years ago.

Unfortunately, recent court proceedings have mocked that promise. Unless some unexpected, bold action restores public confidence, convictions of Chen and his associates will enhance popular cynicism and deny the courts the broad support required by any successful judiciary.

What happened? Chinese have traditionally emphasised substantive criminal law - guilt or innocence - rather than procedure. Yet, recent events, reflecting Taiwan's gradual transition from an inquisitorial to an adversarial system, focused attention on two related sets of criminal process issues: pre-trial detention and the merger of separate prosecutions.

The battle between Taipei District Court Judge Chou Chan-chun's three-judge panel - which twice took the unusual step of ordering Chen's release without bail, pending trial - and Taiwan's High Court - which twice reversed that decision - only ended when the case against the Chen group, originally assigned by lot to Judge Chou's panel, was merged into the earlier prosecution of Chen's wife for embezzling special state funds. That case is being handled by Judge Tsai Shou-hsun's panel.

The transfer made it possible for Judge Tsai to preside over Chen's third post-indictment detention hearing. His panel ordered Chen's return to detention, in a decision that contradicted the spirit of the Council of Grand Justices' Interpretation No653, issued several days earlier. The interpretation eloquently emphasised that the criminally accused should only be detained when no other measures suffice. Although this time Chen is not being handcuffed and held incommunicado, as he was for 32 days before indictment, any conversations with visiting family and even his lawyers can be monitored and used as evidence against him, and them!

Detained defendants are obviously hampered in preparing their defence in other ways, such as by discussing the case with co-defendants and witnesses, which was one of the prosecution's two main fears if Chen remained free pending trial. The other is that, if released, Chen might flee Taiwan. Judge Tsai could have released Chen under high bail and residential restrictions that made flight unlikely. Chen's incentive to flee will increase if he is convicted at trial. Does this mean he will continue to be detained if he appeals against any conviction? This would mean incarceration for years before final conviction.

At what point does the presumption of innocence become meaningless and pre-conviction detention morph into punishment for a crime not finally proved?

The dilemmas of a defendant's detention before final conviction plague every country. More distinctive to Taiwan are the unresolved mysteries surrounding the recent merger of the Chen group's case into the embezzlement case brought against his wife in 2006 - a time when Chen, although involved, still enjoyed presidential immunity from prosecution.

If such a merger was necessary, why was it not effected when the indictment against the Chen group was issued? Instead, the district court decided that the new indictment, which featured money-laundering and other complex charges - plus the earlier embezzlement charge against Chen - should be assigned to a separate judicial panel by lottery.

The lottery was limited to the few panels deemed more specialised than Judge Tsai's for dealing with complex financial transactions. How, then, can the court justify the subsequent assignment to Judge Tsai? Can the random assignment of cases essential to judicial neutrality be so easily circumvented?

According to the court's official press releases, assignment to Judge Tsai by the court's merger review panel followed court rules. Yet that review process could only be initiated by request of the judge in charge of the later case. Why did Judge Chou make that request? Was he pressured to do so? Why did the review panel not accept his proposal to transfer to Judge Tsai only that part of the group indictment relating to the embezzlement case against Chen and his wife, leaving the more complex accusations to Judge Chou's panel, as the court originally intended? Why did the review panel consist of merely five of the relevant criminal division chiefs? Why did the merger issue only become salient after the second time Judge Chou ordered Chen's release? Was this entire non-transparent process the court's response to angry public criticism of Judge Chou? Did any politician intimidate the court with secret threats?

Answers to such questions will eventually emerge. More immediately, is there any way to guarantee the Chen group and the public a judicial process that will have both the appearance and the reality of justice? Why doesn't Judge Tsai, who reportedly did not want to take on the new case, withdraw from handling all but the earlier embezzlement charges against the Chens? Then the district court can return to its original intent and again select by lot a judicial panel of financial specialists to deal with the complex accusations of the new case.

The new panel might even approve another application for Chen's release pending trial, with high bail and strict residential restrictions. Then both trials could proceed with broad public support. Justice, as the saying goes, must not only be done, but must also be seen to be done. That is the price of judicial legitimacy.

Jerome A. Cohen is co-director of NYU's US-Asia Law Institute and adjunct senior fellow at the Council on Foreign Relations