in considering whether there was a real likelihood of bias, the court does not look at the mind of the justice himself or at the mind of the chairman of the tribunal, or whoever it may be, who sits in a judicial capacity. It does not look to see if there was a real likelihood that he would, or did, in fact favour one side at the expense of the other. The court looks at the impression which would be given to other people. Even if he was as impartial as could be, nevertheless if right-minded persons would think that, in the circumstances, there was a real likelihood of bias on his part, then he should not sit. And if he does sit, his decision cannot stand. Nevertheless there must appear to be a real likelihood of bias. Surmise or conjecture is not enough There must be circumstances from which a reasonable man would think it likely or probable that the justice, or chairman, as the case may be, would, or did, favour one side unfairly at the expense of the other. The court will not inquire whether he did, in fact, favour one side unfairly. Suffice it that reasonable people might think he did. The reason is plain enough. Justice must be rooted in confidence: and confidence is destroyed when right-minded people go away thinking: "The judge was biased."
Danckwerts L.J., who dealt with the matter quite shortly, appears to have accepted that it would be enough to justify the court's interference if a person knowing the circumstances might reasonably feel doubts as to the tribunal's impartiality [9] . Edmund Davies L.J. was clearly of the view that the court should interfere if it considered that it would appear to right-thinking people that there were solid grounds for suspecting that a member of the tribunal responsible for the decision may (however unconsciously) have been biased [10] . He expressed his conclusions as follows [11] :
With profound respect to those who have propounded the "real likelihood" test, I take the view that the requirement that justice must manifestly be done operates with undiminished force in cases where bias is alleged and that any development of the law which appears to emasculate that requirement should be strongly resisted. That the different tests, even when applied to the same facts, may lead to different results is illustrated by Reg. v. Barnsley Licensing Justices [12] itself, as Devlin L.J. made clear in the passage I have quoted. But I cannot bring myself to hold that a decision may properly be allowed to stand even although there is reasonable suspicion of bias on the part of one or more members of the adjudicating body.
It has since been doubted whether in practice materially different results would follow from the adoption of one test rather than another: see Hannam v. Bradford Corporation [13] . No doubt in many cases it will be immaterial which test is applied, but that is not universally true, as Edmund Davies L.J. pointed out in the passage already cited from Metropolitan Properties Co. (F.G.C.) Ltd. v. Lannon [14] . It has also been said that the "two tests are often overlapping and it may be that one is appropriate to one situation and another is appropriate to another situation": Reg. v. Altrincham Justices; Ex parte Pennington [15] . However that may be, the judgments of the Court of Appeal in Metropolitan Properties Co. (F.G.C.) Ltd v. Lannon provide authority for the proposition that if there is a reasonable suspicion of bias against a judge that is enough to warrant the court's interference.
1. (1910) 10 C.L.R. 243, at p. 260.
2. (1866) L.R. 1 Q.B. 230, at p. 233.
3. [1894] 1 Q.B. 750, at p. 759.
4. [1955] 1 Q.B. 41, at pp. 47-51.
5. [1960] 2 Q.B. 167, at p. 187.
6. [1969] 1 Q.B. 577.
7. [1924] 1 K.B. 256, at p. 259.
8. [1969] 1 Q.B., at p. 599.
9. [1969] 1 Q.B., at p. 602.
10. [1969] 1 Q.B., at p. 605.
11. [1969] 1 Q.B., at p. 606.
12. [1960] 2 Q.B. 167.
13. [1970] 1 W.L.R. 937, at pp. 942, 949; [1970] 2 All E.R. 690, at p. 700.
14. [1969] 1 Q.B., at p. 606.
15. [1975] 1 Q.B. 549, at pp. 553-554.