JUDGMENT Thomas, C.J. - Before I begin with the facts of the case. I must point out for the future guidance of the learned Sessions Judge and the Government Pleader, that the dates of arrest of the accused persons and the distances from the scene of the occurrence must be brought out in examination-in-chief. The best person to give the dates of arrest is the station officer or the investigating officer but I regret to find that this has not been done in this case and a good deal of time has been wasted in trying to look up the diary etc. Very often the dates of arrest are found in the calendar but this is not good enough, as strictly speaking it is not evidence. * * * * 2. The evidence shows that an armed dacoity took place on the night of the 24th November, 1943, between 10 and 11 p.m. at the house of one Khiman Gadaria, of village Khandaria, a hamlet of Kuchila, police station Sandi, district Hardoi. * * * * 3. Shyam Lal and Dehi Din were put up for identification on the 22nd January, 1943 (vide Ex. 24) and Sheomangal Singh who has been acquitted and Raj Narain were put up on the 7th March, 1944, (vide Ex. 22). There has been considerable delay in the identification proceedings but it seems to me that the accused were not arrested for sometime. Raj Narain was arrested on the 3rd March and Shyam Lal and Debi Din on the 3rd January, 1944. I find in Ex. 24 that three accused were mixed with ten under trials. The proceedings were carried out by Mr. S.P. Bhatnagar, Magistrate first class who was examined on commission as he was transferred to Jaunpur. I have had occasion to criticise his action in Bhup alias Bhupa v. K.E. (Criminal Appeal No. 346 of 1944) decided by a Bench of this Court on the 26th August, 1944. In my opinion the number of under trials mixed was too small. In the case of Emperor Vs. Chhadammi Lal, AIR 1936 All 373 , it was held that in cases in which there is only one or two suspects to be put up for identification, the proportion of one to five cannot be regarded as satisfactory. There should be at least ten under trials for each suspect. In the case of Emperor Vs. Chhadammi Lal, AIR 1936 All 373 , it was held that in cases in which there is only one or two suspects to be put up for identification, the proportion of one to five cannot be regarded as satisfactory. There should be at least ten under trials for each suspect. It will of course depend upon the circumstances of each case because sometimes it may not be possible to get a proportion of ten to one particularly where there is a large number of accused. In the case of Ram Singh v. K.E. 1943 O.A. 49 : A.W.R. (C.C.) 21 : O.W.N. 88 it was held by a Bench of this Court, of which I was a member, that where the proportion of other undertrial prisoners to the accused is less than three to one, identification have little value because there is appreciable risk of persons being implicated purely by chance. It is true that the law does not lay down the number of undertrials to be mixed with a prisoner, but every effort should be made to minimise the possibility of a chance identification which in the first instance can be done by mixing as many undertrials as possible with the suspect. I have not the slightest doubt that a large number of undertrials would have been available in this case and if the Magistrate had taken the trouble this difficulty would not have arisen. 4. The second identification (Ex. 22) was carried out by Mr. Sripat, Magistrate first class. In this case ten undertrials were mixed with two accused. This by no means can be said to be a large number but certainly is better than Ex. 24. 5. I do not think it necessary to deal separately with the case of each accused named in the first information report. * * * * 6. The result is that I allow the appeal of Sham Lal, Raj Narain and Manohar and direct that they be released forthwith. I dismiss the appeal of Rameshar, Karhley, Dirvijai Singh, Jagannath Singh, Chotey Lal, Dila Ram, and Himanchal.