THE COURT OF CRIMINAL APPEAL
[C.C.A. No. 116CPA of 2003]
IN THE MATTER OF AN APPLICATION PURSUANT TO SECTION 2 OF THE CRIMINAL PROCEDURE ACT, 1993
MICHAEL JOSEPH KELLY
THE DIRECTOR OF PUBLIC PROSECUTIONS
JUDGMENT of the court delivered on the 1st day of February 2008 by Kearns J.
Michael Joseph Kelly, the applicant herein, was convicted on 24th March, 1983, by a jury in the Central Criminal Court (Gannon J.) of the murder of Margaret Glynn at the house which she occupied with her elderly brother Martin Glynn, at Keeves, Ballinamore Bridge, Co. Galway on 15th November, 1981.
Leave to appeal was refused by this Court on 25th May, 1984, and the applicant was later refused a certificate to appeal to the Supreme Court pursuant to s. 29 of the Courts of Justice Act, 1924.
Following conviction the applicant was sentenced to penal servitude for life. He was however released on licence in the mid 1990s.
The applicant now seeks to have his conviction for murder quashed pursuant to the provisions of s. 2 of the Criminal Procedure Act, 1993 on the basis that there are newly discovered facts which show that there has been a miscarriage of justice in his case.
THE CRIMINAL PROCEDURE ACT, 1993
The Criminal Procedure Act, 1993 makes provision for the Court of Criminal Appeal to review alleged miscarriages of justice in cases where the court has previously rejected an appeal or an application for leave to appeal in the case.
Section 2 of the Act provides as follows:-
In the instant case, the application is grounded on s. 2(4), that is to say it is based on what are contended to be “newly discovered facts” coming to the notice of the applicant and his advisors after his appeal herein had been disposed of.
“(1) A person –
(a) who has been convicted of an offence either –
(b) who alleges that a new or newly discovered fact shows that there has been a miscarriage of justice in relation to the conviction or that the sentence imposed is excessive,
(i) on indictment, or
(ii) after signing a plea of guilty and being sent forward for sentence under s. 13 (2) (b) of The Criminal Procedure Act, 1967, and
who after appeal to the Court including an application for leave to appeal, and any subsequent re-trial, stands convicted of an offence to which this paragraph applies, and
may, if no further proceedings are pending in relation to the appeal, apply to the Court for an order quashing the conviction or reviewing the sentence.
(2) An application under subsection (1) shall be treated for all purposes as an appeal to the Court against the conviction or sentence
(3) In subsection (1) (b) the reference to a new fact is to a fact known to the convicted person at the time of the trial or appeal proceedings the significance of which was appreciated by him, where he alleges that there is a reasonable explanation for his failure to adduce evidence of that fact.
(4) The reference in subsection (1) (b) to a newly-discovered fact is to a fact discovered by or coming to the notice of the convicted person after the relevant appeal proceedings have been finally determined or a fact the significance of which was not appreciated by the convicted person or his advisers during the trial or appeal proceedings.
(a) after an application by a convicted person under subsection (1) and any subsequent re-trial the person stands convicted of an offence, and
(b) the person alleges that a fact discovered by him or coming to his notice after the hearing of the application and any subsequent re-trial or a fact the significance of which was not appreciated by him or his advisers during the hearing of the application and any subsequent re-trial shows that there has been a miscarriage of justice in relation to the conviction, or that the sentence was excessive,
he may apply to the Court for an order quashing the conviction or reviewing the sentence and his application shall be treated as if it were an application under that subsection.”
Before considering the elements of what may constitute a “newly discovered fact” it is necessary to set out the background to this case.
In the early morning of 15th November, 1981, a fire occurred in a three-roomed cottage occupied by Martin and Margaret Glynn in a rural area near Ballinasloe, Co. Galway. The Glynns were brother and sister, both of whom were elderly and unmarried (Martin Glynn was aged 85 years and Margaret Glynn aged 87 years). Martin Glynn was in poor health and unable to look after himself but his sister was in somewhat better health. At the time of their death, both were being looked after by the applicant, who at the time was aged 29 years and resided with them, and who was staying in the house on the same night.
Following the fire, Margaret and Martin Glynn were found dead in the house and brought to Portiuncula Hospital in Ballinasloe. The applicant was interviewed at Ballinasloe Garda Station later that day on Sunday, 15th November, 1981 when he outlined how he had woken up to find the house on fire and that he could not save either Martin or Margaret Glynn. In his initial statement, the applicant lied in relation to the time he was in the house, as the Glynns - following a deterioration of the relationship between them and the applicant - had locked him out the previous day and he had had to force open the door in the hours preceding the fire in order to gain entrance to the house.
A post mortem was carried out on the bodies of Martin and Margaret Glynn on 16th November, 1981 commencing shortly before 3pm by the then State Pathologist, Dr. John Harbison. Dr. Harbison concluded that Martin Glynn had died of natural causes, namely bronchopneumonia. He was also of the opinion that Martin Glynn had died prior to the commencement of the fire.
In relation to Margaret Glynn, Dr. Harbison noted that the body was clothed and that some of the clothing was burned. The body was severely burned in places, particularly the left side of the head and neck. There was bruising on the under surface of the scalp on the right side which extended down the right cheek bone and into the tissues on the right side of the neck. Further bruising was identified in the muscles overlying the voice box and below the thyroid gland in the front of the neck. A fracture of the thyroid cartilage was suspected. There was no evidence of soot inhalation in the air passages and an analysis of a sample of blood taken from the deceased revealed a carboxyhaemoglobin concentration of less than 2%.
Dr. Harbison concluded that Margaret Glynn’s neck injuries were indicative of strangulation or throttling. He was of the opinion that she was dead when the fire started and that the distribution of the burns raised the possibility that the fire was a deliberate attempt to conceal homicide.
While at trial Dr. Harbison accepted that there was not in fact a fracture of the larynx, he was nonetheless satisfied that death in the case of Margaret Glynn was due to strangulation.
On 22nd November, 1981 the applicant was arrested and detained at Ballinasloe Garda Station pursuant to the provisions of s. 30 of the Offences Against the State Act, 1939. On the evening of 22nd, the applicant signed a written statement in which he admitted to the murder of Margaret Glynn.
He described how on the day of 14th November he had sought admission to the Glynn’s house but had been told to go away. The applicant tried repeatedly to gain access, but without success. At one stage he forced open the door which had been held closed by a stick. Having repaired thereafter to a neighbour’s house, the applicant again went up to the Glynn’s house during the small hours and lay down on a bed which he had in the house until it was daylight. His statement then continued:-
The applicant was charged with murder and arson contrary to s. 2 of the Malicious Damage Act, 1861 and his case was heard before Mr. Justice Gannon and a jury in the Central Criminal Court between 21st day of March, 1983 and the 24th day of March, 1983. The trial judge directed the jury to find the applicant not guilty of arson. The applicant was found guilty of murder and sentenced to penal servitude for life.
“I got up then and I went over to Marty’s bed and I saw that he was not breathing. I then knew that he was dead. Maggie had me driven daft and I went down to her bedroom. It was bright at this stage and I saw the candle lighting beside her bed on a chair. I went to her bed and I saw that she was asleep. I picked up a blanket off her bed and I put it over her face. I then pressed on her neck with my right hand and she started shouting ‘don’t do it Micheleen’. I kept pressing on her neck until she stopped breathing. I then lifted the blanket off her head and I held it over the candle until it caught fire. I then held the burning blanket to the outside of the bed until it caught fire. When the outside of the bed was on fire I threw the burning blanket on the inside of the bed near the wall. The top part of the bed went on fire and the room was soon all smoke. I went out then from the house and went down to Mickey Donoghues.”
The case against the applicant at his trial was based on two main pillars. Firstly, there were the findings of the pathologist who carried out the post mortem on both Margaret and Martin Glynn. Dr. Harbison’s view was that Margaret Glynn died from asphyxia caused by compression of her neck in view of injuries which he found, including bruising of her neck, a fracture of her larynx and bleeding into her lungs. Secondly, the prosecution relied on the statement made by the applicant in Ballinasloe Garda Station in which he admitted to killing Margaret Glynn. This confession was made by the applicant to Detective Garda Joseph Shelly and Sergeant Patrick Lynam, both from the Garda Technical Bureau. The prosecution also relied to some extent on evidence that the applicant had asked Mr Donoghue on the night of the 15th November to tell the gardai nothing about his movements or whereabouts the previous evening. Reliance was also placed on a further conversation which the applicant had some three days prior to the deaths with a local man, Mr Thomas Geraghty, in which the applicant expressed his concerns that if Margaret Glynn survived her brother, she would sell up the cottage and he would be forced to leave with nothing, despite having been told by the Glynns that the place was to be left to him by way of recompense for looking after them while they were alive.
The applicant’s defence at trial was that he had no hand, act or part in the arson of the house at Keeves, Ballinamore, Co. Galway, nor had he anything to do with the killing of Margaret Glynn. The medical evidence relied on by the prosecution was challenged both through cross-examination and by the evidence of an expert, Dr. Declan Gilsenan, a pathologist, who was called by the defence and who disagreed with some of the findings of Dr. Harbison. The applicant gave evidence in a voir dire hearing that his statement was not made voluntarily but rather was made only following intimidation by the gardaí who had interviewed him. However, the learned trial judge ruled that the said statement had been made voluntarily, and the same was duly considered by the jury as part of the evidence in the case. At trial, the applicant made an unsworn statement before the jury denying any involvement in the murder of Margaret Glynn.
The application originally contained two grounds of appeal raising issues of alleged newly discovered facts relating, firstly, to the pathology evidence and, secondly, to the authenticity of the inculpatory statement made by the applicant. In relation to the latter it was alleged that technology which was not available at time of trial, namely the CUSUM test, suggested that there had been more than one contributor to the making of the statement so that it could no longer be safely regarded as the applicant’s own statement.
By way of further motion and grounding affidavit dated 9th December, 2005, an application was made to add further grounds of appeal. One of these grounds sought to rely on certain information given to Dr. Harbison prior to the autopsy which was not disclosed to the defence at time of trial. This was not information of a medical nature, but information which suggested a motive on the part of the applicant for the murder of Margaret Glynn. It was suggested that this might have influenced the views of Dr. Harbison and may have prejudiced him in his performance of his duties as pathologist in this instance. This ground was not, however, pursued before this Court.
The further grounds of appeal may be summarised as follows:-
By notice of motion dated 12th November, 2007, the applicant’s solicitors sought leave to add a further ground of appeal following the discovery by the applicant’s solicitor of post mortem photographs depicting the dissection of the neck of the late Margaret Glynn deceased as carried out by Dr. Harbison, which said photographs were stated not to have been available to the defence at trial. These photographs had since been viewed on behalf of the applicant by Dr. Basil Purdue, Home Office Pathologist, who concluded that these photographs made it plain that the body of Margaret Glynn was severely affected by both heat involving the left side of the neck and autolysis of such a degree that bruising could not be safely diagnosed.
(a) The conclusion drawn from the findings that the larynx of the deceased was flexible and misshapen as being evidence consistent only with death having been caused by manual strangulation could no longer be safely relied upon having regard to the opinions of a number of medical experts
(b) The finding that there were petechiae in the lungs of the deceased, which was treated as evidence of death having been caused by manual strangulation, could no longer be relied upon as safe, as so indicated by the opinion evidence of various medical experts
(c) There was a failure to have regard to the reasonable possibility that the death of the deceased was caused by the inhalation of cyanide gas occasioned by the release of gas in the burning of a pillow on the bed of the deceased, as indicated in the further opinion evidence of a number of technical and medical experts
(d) The prosecution theory that the fire started on top of the bed of Margaret Glynn as claimed by the applicant and as stated by him in his alleged statement to the Gardaí was incorrect having regard to further expert evidence from a fire consultant, whose opinion was to the effect that the fire started beneath the bed
It is important to set out the background to the emergence of these photographs as deposed to by Gregory F. O’ Neill, the applicant’s solicitor, in his affidavit supporting the notice of motion. He deposes that the applicant first consulted him in 1995, but the applicant had little or no papers concerning his case apart from some “vestigial” documents pertaining to his grounds of appeal. Mr. O’ Neill made inquiries of his previous solicitor, who unfortunately had become ill and had not retained any papers. However, through the assistance of the Registrar of the Court of Criminal Appeal, Mr. O’ Neill obtained the file held in the office of the Central Criminal Court, which included a copy of the Book of Evidence and some other documents, including two booklets of photographs. He also retrieved from the Office of the Attorney General material pertaining to an application unsuccessfully brought by the applicant when he sought a certificate pursuant to s. 29 of the Courts of Justice Act, 1924.
Ultimately, Mr. O’ Neill obtained a copy of the trial transcript in 1999. On 2nd July, 2004 the office of the Chief Prosecution Solicitor sent to Mr. O’ Neill a bundle of documentation which included further copies of the two booklets of photographs already referred to and a third booklet of photographs. The first booklet of photographs contained photographs numbered 1 to 12. The second book of photographs contained photographs numbered 13 to 16. The third book of photographs was unnumbered.
This third bundle of photographs constituted a contemporaneous photographic record of Dr. Harbison’s examination and dissection of the remains of the late Margaret Glynn. Although this book of photographs had been sent to a firm of solicitors with whom Mr. O’ Neill had been employed in 2004, he did not actually receive the material until some time later. When he examined the Book of Evidence and notices of additional evidence, it became plain that only two albums of photographs had been supplied to the defence prior to the trial in 1983.
Having obtained the material which included this third booklet of photographs, it never occurred to Mr. O’ Neill that the photographic material contained material which was additional to the two original booklets of photographs which he had obtained some years previously. Accordingly, when he first briefed Dr. Basil Purdue, Home Office Pathologist it was only with the first two sets of photographs. The third booklet was however available when a consultation with Dr. Purdue took place at his offices on 1st November, 2007. This was the first occasion when any expert retained by the defence had an opportunity of looking at the contemporaneous photographic record of the post mortem examination carried out by Dr. Harbison on the remains of the late Margaret Glynn.
From further enquiries then undertaken by Mr. O’ Neill it transpired that Dr. Jack Crane, a Consultant Pathologist retained by the Chief Prosecution Solicitor in connection with this appeal, had equally not been furnished with the particular booklet of photographs. This was immediately rectified so that Dr. Purdue and Dr. Crane were subsequently in a position to discuss the implications of the new photographic material.
The motion was also supported by an affidavit from Mr Patrick McEntee, the distinguished Senior Counsel, who had been the applicant’s counsel at trial, in which he confirmed he had no recollection of having been furnished with the photographs in the third booklet at the time of the trial. His recollection is buttressed by the complete absence of any reference to these photographs in the transcript of the trial.
No objection having been raised by Mr. O’ Connell, Senior Counsel for the respondent, the court directed that this further ground of appeal be added to those already lodged.
WHAT IS A ‘NEWLY DISCOVERED FACT’ ?
By way of preliminary issue, Mr. O’ Connell argued that neither the various newly - commissioned expert reports nor the opinions expressed therein could be seen as constituting ‘newly discovered facts’ within the meaning of the Criminal Procedure Act, 1993. They were merely further and additional opinions, albeit opinions informed by advances in science. In the present case, with the exception of the photographs and CUSUM analysis, there were no new facts and no new science such as might invalidate any opinion tendered at trial by Dr. Harbison. He submitted that the procedure under section 2 of the Act does not introduce an automatic right to a hearing merely because an expert report can be produced contradicting the prosecution expert at the original trial. Such an application may constitute no more than an attempt to introduce a fresh expert who might do a better job than the original defence expert. Alternatively, there could be a broad brush/ approach in the course of which an applicant could seek to introduce a mixture of old and new science and/or old and new grounds of appeal in such a way that a limited piece of “new science” could be used as a trigger for a complete rehearing of numerous other issues.
Specifically in relation to the various grounds of appeal advanced, Mr. O’ Connell addressed the areas of opinion evidence in the following manner:-
It was conceded by Dr. Harbison at the original trial in 1983 that there was no fracture of the larynx although he had initially thought this to be the case. This issue was canvassed at trial and was clearly not a newly discovered fact.
(a) The supposed fracture of the larynx.
Mr. O’ Connell again submitted that none of the reports later commissioned on behalf of the applicant suggested the availability of some new science that was not available in 1983 at the time of the trial, nor had the existence of any existing science rendered invalid the findings at trial.
(b) Patchy bleeding into the lungs of Margaret Glynn
Mr. O’ Connell accepted that the reports of Dr. Purdue (on behalf of the applicant) and Dr. Crane (on behalf of the respondent) suggest that testing for cyanide poisoning has become more routine in modern times. He also accepted that the report of Professor Mayes, Dr. Purdue and Dr. Forrest (all experts retained on behalf of the applicant) referred to modern studies of cases where the possibility of cyanide poisoning unaccompanied by increased carbon monoxide or soot in the airways (as was the case with Margaret Glynn) may explain a death. This information was not available in 1983, but Mr. O’ Connell pointed out that the remains of Margaret Glynn had not been actually tested for cyanide poisoning so that any opinion evidence in this respect would be purely speculative.
(c) The possibility of a cyanide poisoning of the deceased
Mr. O’ Connell submitted that although the fire consultant retained on behalf of the applicant issued his report in 2006, it was not clear that it contained any expertise about fires that was not known in 1983. He thus submitted that opinion evidence as to where the fire might have started, which was not called by the defence at trial, should not now be regarded as being admissible as a newly discovered fact. Any theory advanced by any such expert that the fire may have caused the death and/or that the fire might have started under the bed could be discounted on evidence given at trial to the effect that an inspection of the underneath of the bed in 1981 revealed no hole such as might be caused by burning.
(d) Evidence of fire expert suggesting the fire began under the bed
Anthony Sammon, Senior Counsel on behalf of the applicant, accepted that there was a difficult line to be drawn between opinion and fact. If new scientific opinion offered a more reliable guide to what might have happened in any given situation, he submitted that the court should treat it as evidence of a newly discovered fact. He accepted, however, that the remains of Margaret Glynn had not been tested for cyanide poisoning, but argued that this was now an unexplored possibility in the case which raised a doubt and that the applicant should have the benefit of it. Mr. Sammon further accepted that a review of the various Irish authorities did not indicate any precedent for holding that opinion evidence could be regarded as constituting a newly discovered fact. He argued however that commonsense would suggest that where scientific developments offered a more reliable basis for the expression of an expert opinion at a later stage, then evidence of such opinion should be admissible.
It is a basic principle of statutory interpretation that words be given their natural and ordinary meaning. The Concise Oxford English Dictionary contains the following useful definition of the word ‘fact’: “A thing that is indisputably the case”; “the truth about events as opposed to interpretation”. The word ‘opinion’ on the other hand is defined as: “a view or judgement not necessarily based on fact or knowledge.”
It seems to the Court that there is a critical difference of meaning evident in these definitions, an impression that is fortified by reference to decided cases in this jurisdiction wherein evidence of “newly discovered facts” was admitted. In The People (Director of Public Prosecutions) v. Meleady & Grogan  2 IR 517 the “newly discovered fact” was evidence of a fingerprint found on the inside of a front passenger door window in a car. The gentleman whose fingerprint it was gave evidence on behalf of the applicants at the trial that he had been the front seat passenger in the vehicle that night and that neither of the applicants had been in the car which had been stolen by a number of youths. The owner of the car had attempted to stop the youths by jumping on the bonnet of his car and had purported to identify the first applicant as the driver of the car and the second applicant as the front seat passenger. Following a further conviction after a retrial, it emerged for the first time that the fingerprint of a person other than either applicant had been found on the inside of the front passenger door window. This was clearly a “newly discovered fact”. An inquiry which was then undertaken in the Chief State’s Solicitor’s Office by a Principal Solicitor in that office led to the discovery of a written memorandum prepared by R.W., solicitor in the District Court section of the office (the “Walker Memorandum”) which was addressed to the Circuit Court section of the office and referred to a particular garda having mentioned that the main prosecution witness, the owner of the motor car, had been shown a book of photographs and had identified one of the applicants therefrom. The particulars of that case need not be further considered for present purposes other than to state that the two matters which persuaded this Court to receive evidence were clearly factual matters and not matters of opinion.
In The People (Director of Public Prosecutions) v. Pringle  2 IR 547 this Court considered the following as capable of amounting to “newly discovered facts”:-
While the court ultimately resolved the particular case by reference to the conflict which existed between the two police officers and made no finding about the reliability or otherwise of the CUSUM technique, it is nonetheless clear that all matters in respect of which the court heard evidence were matters of fact and not opinion.
(a) Evidence of the CUSUM technique, assuming if admitted that it would have raised a doubt in the mind of the jury
(b) Evidence that a tissue had not been forwarded to the State Forensic Laboratory for analysis so that the court of trial was deprived of the results of an analysis which might have been of assistance in establishing the innocence of the accused
(c) The non-disclosure of the circumstances surrounding the forwarding, or not forwarding, of the tissue to the State Forensic Laboratory and the non-disclosure of the conflict between two members of the gardai as to what happened the tissue
In The People (Director of Public Prosecutions) v. Gannon  1 I.R. 40, two documents came to light, copies of which had not been furnished to the applicant’s legal advisors at the time of his trial. The first consisted of notes taken by a guidance counsellor who was the first person to whom the complainant had reported a rape, and the second was a report which a garda had prepared based upon those notes which she had received from the counsellor. Both the notes and the garda report contained details of the description given to the counsellor by the complainant of the man who raped her. Clearly in this case also the newly discovered material was factual in nature.
In the People (Director of Public Prosecutions) v McDonagh  1 I.R. 305, the ‘newly discovered fact’ evidence consisted of a signed statement brought into existence subsequent to the appeal in this Court in which the applicant’s co-accused sought to exonerate the applicant and accept sole responsibility for a sexual assault – again plainly a factual matter only.
More recently in The People (Director of Public Prosecutions) v Shortt (No. 1)  2 I.R. 686, the non-disclosure prior to trial of serious allegations raised against the principal State witness were conceded by the State to constitute newly discovered fact within the meaning of section 2 of the Act of 1993. Again, no issue of opinion evidence arose in this case either.
In ruling on the present application the Court is strongly of the view that opinion evidence, subject to the qualification hereinafter expressed, should not constitute newly discovered fact within the terms of the Criminal Procedure Act, 1993. Firstly, to so interpret opinion evidence would be to give a meaning to the word “fact” which is quite different from its ordinary and natural meaning. Secondly, it would have the effect of rendering virtually every conviction, even one upheld by this court following an appeal, open to later challenge if a further or new expert could be found to offer an opinion which went further than a defence expert had done at trial, or which tended to contradict or undermine experts called on behalf of the prosecution at trial. It would open the door to the introduction of additional evidence in circumstances which were plainly contra-indicated by this Court in The People (Director of Public Prosecutions) v Willoughby  IECCA 4. Having conducted an extensive review of both the Irish authorities and a number of English authorities in that case, this Court considered it could formulate principles appropriate to an application to introduce new or fresh evidence in the Court of Criminal Appeal as follows (at pp. 21-22):-
This approach to the admissibility of new evidence on appeal was expressly endorsed by the Supreme Court in The People (Director of Public Prosecutions) v. O’ Regan  IESC 38 and must now be seen as settled law in this respect.
“(a) Given that the public interest requires that a defendant bring forward his entire case at trial, exceptional circumstances must be established before the court should allow further evidence to be called. That onus is particularly heavy in the case of expert testimony, having regard to the availability generally of expertise from multiple sources
(b) The evidence must not have been known at the time of the trial and must be such that it could not reasonably have been known or acquired at the time of the trial
(c) It must be evidence which is credible and which might have a material and important influence on the result of the case
(d) The assessment of credibility or materiality must be conducted by reference to the other evidence at the trial and not in isolation.”
It would in my view be altogether impermissible for this Court to adopt an approach to opinion evidence which both ignores the express terminology of s. 2 of the Criminal Procedure Act, 1993 and also, by implication at least, goes totally against the thrust of the two decisions to which I have just referred.
That is not to say that opinion evidence is in all circumstances inadmissible, as the Court’s present ruling will make clear. There may be cases where a state of scientific knowledge as of the date of trial may be invalidated or thrown into significant uncertainty by newly developed science. There may also be cases where the opinion of an expert at trial may later be shown to have been tainted by dishonesty, incompetence or bias to such a degree as to render his evidence worthless or unreliable. Once such ‘facts’ are established, expert opinion evidence must clearly then be admissible so that such new ‘facts’ can be properly interpreted.
It is perhaps easiest to illustrate the distinction between opinion and fact by reference to the evidence which the Court does propose to consider in this matter. The Court is quite satisfied that the evidence contained in the third booklet of photographs, and expert evidence in relation to the interpretation of these photographs, is admissible as “newly discovered fact”. These photographs were not available to the defence either at the time of trial, or the appeal. The emergence and existence of these photographs is quite obviously a factual matter and capable of being established as such in evidence in the sense envisaged by the Act.
Equally, the development of the CUSUM technique, which in essence is directed to ascertaining whether more than one contributor was involved in the making of a statement, is a science which, though no longer new, was not available at time of trial. The Court is disposed to hear evidence about the technique, and will of course also entertain opinion evidence as to its reliability and credibility. It will also hear opinion evidence as to the interpretation of any analysis undertaken by reference to the CUSUM test.
The Court however will decline to receive evidence on the remaining matters which, in the opinion of the Court, are entirely matters of expert opinion falling short of the criteria outlined above.
Dr. Basil Purdue is an independent forensic pathologist since 1978 who has been in full-time practice of forensic medicine and pathology since 1981. He is on a list of pathologists supplied by the Home Office who provide a service to the police forces of Hampshire, Wiltshire, Dorset and Avon and Somerset and to solicitors in the South and West of England. He holds the qualifications of Bachelor of Science and Bachelor of Medicine and Surgery. He is a fellow of the Royal College of Pathologists and holds a Diploma in Medical Jurisprudence (Pathology) from the Society of Apothecaries of London. He is also joint author of a well known textbook “The Pathology of Trauma” with Professor J.K. Mason.
He told the Court in evidence that he had originally obtained two booklets of photographs, but in recent days had received and examined the third booklet which was of high quality although black and white.
He gave evidence that in the photographs one could see evidence of heat haematoma typical of the effects of heat on a dead body. He stated that heat sends the body into decay more quickly than usual. He noted that Dr. Harbison had only examined the corpse after a significant period of time had elapsed. All of the photographs showed severe decay and in his opinion there was no reliable evidence of bruising from the photographs. There was evidence of severe autolysis which made it unsafe to diagnose bruising in the neck area.
He stated that the pathological portion of the prosecution case rested on three pillars. The first was the suggestion of a fractured thyroid, a suggestion Dr. Harbison had withdrawn at trial. The second was evidence of patchy haemorrhage in the lungs, but this was non specific in nature and could not safely be relied upon. The third pillar of the prosecution case consisted of bruising of the neck, and in Dr. Purdue’s opinion, none of the bruises described by Dr. Harbison could be reliably identified anywhere in the booklet of photographs. The furthest it could be put was that there was evidence of discolouration which may or may not have been bruising. He felt the cause of death should have been recorded as “unascertainable”.
Professor John Crane is State Pathologist for Northern Ireland. He is also Professor of Forensic Medicine at Queens University and a Consultant in Pathology to the Northern Ireland Health & Social Services Boards. He also had access to all the papers in the case, including the post mortem reports of Dr. Harbison in respect of his autopsies on Martin Glynn and Margaret Glynn, garda photographs of the post mortem examination of Margaret Glynn and histology slides prepared by Dr. Harbison in respect of both autopsies.
Before seeing the lately discovered photographs, Professor Crane accepted that Dr. Harbison did incorrectly diagnose a laryngeal fracture, but pointed out that Dr. Harbison had accepted that fact in the course of the trial but nonetheless still remained of the opinion that death was due to strangulation. When reporting in August, 2004, Professor Crane was of the view that a pathologist with the experience of Dr. Harbison would not misinterpret any bruising he found as being due to heat necrosis. He felt that the post mortem photographs were of little assistance in confirming the nature and extent of the bruising and that Dr. Harbison had not taken any sections for microscopy to confirm its nature. He felt that the injuries described by Dr. Harbison could be consistent with the application of pressure on the neck and probably also as a result of blows to the scalp and face. He was also of the view at that time that it was difficult to say that the cause of the injuries was manual strangulation, because the absence of fractures of the hyoid and laryngeal cartilages and the failure to find petechial haemorrhages in the right eye would mitigate against such a finding. However, in his experience the extent of bruising in such cases is relatively slight and quite subtle. However, it was not unreasonable to conclude that whatever the mechanism of injury to the neck it could have been sufficient to account for death. He was also in agreement with Dr. Harbison that the severity of the burns to Margaret Glynn would be in keeping with a deliberate attempt to burn the body to conceal injuries responsible for her death.
In the course of his evidence Professor Crane said that the photographs of the clothed body of Margaret Glynn in the mortuary show evidence of severe burns in the form of charring to the left side of deceased’s face, neck and left shoulder/upper arm. It appears that the left side of the deceased’s head was close to the seat of the fire, given that she was lying on her left side in the bed.
Having looked at the new photographs, in particular those showing areas of discolouration, he felt unable to confirm that they – taken by themselves alone - demonstrate evidence of bruising. That, however, was not to say that bruising was not present, but simply that it was not demonstrated in the photographs. He accepted that, in tissues undergoing decomposition, discolouration may develop which may mimic bruising. Thus there was a possibility that discolouration due to post mortem change was misinterpreted by Professor Harbison as evidence of antemortem injury.
Professor Crane stressed however that there was no evidence of death by natural causes. Further, there was no evidence to indicate that Margaret Glynn was alive when the fire started. Had she been alive at that stage, she would have inhaled smoke thereby drawing soot into her airways and carbon monoxide into her lungs. However, no soot had been found in her airways and the 2% level of carbon monoxide found to be present in her blood was of no significance and did not indicate smoke inhalation.
In this case there had been what Dr. Harbison believed to have been bruising in the parietal area of the head on the right side, over the right side of the jaw, and in the neck area. This was apparent from internal examination of blood vessels.
While there was no unequivocal evidence of strangulation, the discolouration found and noted by Dr. Harbison was localised to the right side of the head and neck and was not uniform. If the discolouration was due to putrefaction, it was more likely to be uniform. Furthermore, post mortem staining was more likely to be affected by gravity, so that in interpreting the findings made by Dr. Harbison, it had to be remembered that the body was found lying on its left side, whereas the discolouration or bruising in this case was to the right side of the head, jaw and neck. This suggested to Professor Crane that the discolouration was more likely to be bruising.
In his opinion, the black and white photographs were of limited value to the defence. They were internal photographs and for that reason extremely difficult to interpret.
In cross examination, Professor Crane dismissed any suggestion that the cause of death might have been due to any disease associated with the fact that the lymph glands of the deceased were somewhat abnormal. While he accepted that he would not have used the words “manual strangulation” or “throttling” to describe the cause of death, he would nonetheless have ascribed the death to compression of the neck of Margaret Glynn. He stressed more than once that there was no evidence of any natural disease to explain the death.
Evidence in relation to CUSUM Technique
Dr. Michael Farringdon is a chartered engineer. He is a Bachelor of Science in Electronic Engineering and has a Master’s Degree in English Language & Literature. He is also a Doctor of Philosophy in computer science. His expertise lies in linguistic analysis and he is mainly associated with the development of CUSUM analysis which was pioneered by Rev A.Q. Morton, a computer scientist from Glasgow. He has given evidence in a number of criminal trials in this jurisdiction, although the last occasion he did so appears to have been in 1996. The majority of the cases were in 1991 to 1992.
Dr. Farringdon gave evidence of how CUSUM analysis employing the statistical technique of cumulative sum analysis operates to determine the origin of human utterance in either of its two common forms, the written word or the spoken word. The decision about the authorship is made by comparing two graphs. These are a simple type, known as cumulative sum charts, so giving the name to the technique.
In CUSUM analysis it is usually the case that the sentence is chosen as a unit of measure. The sentence is usually not so long that order and arrangement in the word classes being counted are diluted, nor does one usually have a large number of sentences so short that there is little information left in them. Some editing may take place to improve the punctuation. This might involve conflating compound names and by omitting or conflating lists, by omitting very short sentences of one, two, three or four words, by omitting repetition, and by omitting answers which largely repeat words in the preceding question.
One then looks to observe the use of habit words by the subject and draw a comparison between the instances of habit words in known utterances of the subject with disputed material. Only the most frequent habits can be expected to show any positive result. An examination of two, three and four letter words and words beginning with a vowel are typically reflected in a graph. The decision about authorship is then made by comparing two graphs. The first graph shows an aspect of all the sentences in the sample and is based on the average number of words per sentence and shows how the sentences in the sequence vary around that average. A second graph shows how some component part has been used to construct the sentences. Thus a graph can show how nouns or other syntactic have been used to make sentences. In the speech or writing of one person the two graphs will track each other closely.
In the instant case the documents analysed were the transcript of the spoken evidence of the applicant given on the 23rd March, 1983. Two controlled samples of utterance were taken from this document, consisting of one segment of 56 sentences and a second segment of 29 sentences. Both samples were stated to show a consistent habit running through the sample of utterance.
The two controlled samples were then combined into a single sample and this combined sample of 85 sentences was tested for homogeneity and again showed a consistent habit.
Dr. Farringdon then applied a weighted CUSUM factor to obtain probabilities of consistency of habit. He explained that the work of Professor A. F. Bissell ‘Statistical Methods for Text Analysis by Word Counts’ (Swansea: European Business Management School, University of Wales, 1995) provided the basis for this exercise. In brief, this approach means that while all sentences are given equal weight on conventional CUSUM charts, a long sentence is given more weight than a short sentence on the weighted CUSUM chart.
A probability of less than 0.1 indicates that the sample requires further investigation; a probability of less than 0.05 is significant, less than 0.01 very significant, and less than 0.005 highly significant.
Having compared the applicants controlled sample (i.e. his evidence in court) with the confession, Dr. Farringdon gave evidence that the combined sample showed a clear separation of the two graphs indicating that the confession was the output of two or more authors. The weighted CUSUM chart showed a similar deviation, producing an adjusted probability of less than 0.005 which was, in his opinion, highly significant. In Dr. Farringdon’s opinion, the alleged voluntary and dictated confession of the applicant was the mixed utterance of two or more people.
Dr. Farringdon was cross examined vigorously by Mr. O’Connell who suggested that the “so called science of CUSUM technology” had no established scientific validity. Dr. Farringdon accepted that evidence of this nature had been challenged on various occasions when he had endeavoured to give evidence and that it had never been accepted by any court in the United States of America in the context of a criminal trial. Dr. Farringdon accepted that CUSUM was a “minority interest” which was perhaps most usefully deployed in literary detective work, such as endeavours to ascertain if the works of Shakespeare involved more than one author. He himself had applied CUSUM analysis to the works of Henry Fielding.
Asked if he could point to any body of scientific peer review supporting the accuracy of the technique, Dr. Farringdon mentioned his own published work, mostly in the 1980’s, but referred also to a contribution made by him to a work on the topic written by his wife Jill Farringdon in 1996. There was also one publication in relation to the technique in the United States of America, but he accepted that the paper in question was one prepared by his wife who, with him, was co-author of their published work on this topic.
He accepted that editing of text had a degree of subjective input but did not agree that this posed any threat to the integrity of the technique.
Asked by the Court if he was now willing to analyse a blind sample, Dr. Farringdon expressed considerable reservations, stating that this would involve considerable time and expense. It was put to him that the proponents of CUSUM have never reported carrying out such a controlled or blind trial and that Dr. Farringdon had previously refused to co-operate in one such trial with the respondent’s expert, Professor Canter, but insisted that a blind, controlled study was not feasible.
In fairness to Dr. Farringdon, he insisted that the reason for not proceeding with the blind trial in that particular instance was because he and Professor Canter could not agree ground rules for the study.
The report of Professor David Canter was admitted as evidence for the respondent by agreement between the parties.
Professor Canter is Professor of Psychology at the University of Liverpool. His main area of expertise is the systematic analysis of human behaviour in order to identify dominant trends within it. He has made a particular study of procedures that claim to indicate the authorship of contested material. He has written numerous studies of the CUSUM technique which have been published in learned journals.
In summary, he concluded that the CUSUM procedure used by Dr. Farringdon has absolutely no established scientific validity. In his opinion the procedure cannot be relied upon to give any indication whatsoever of the authorship, single or otherwise, of the utterances attributed to the applicant.
He explains that examination of authorship “styles” has been the subject of scholarly debate for over a century. However, all that has been established is that if very large samples of a person’s writings are compared with very large samples of the writings of another person, variations in features of those two set of writings can sometimes be established.
He described a continuum of opinion in relation to the debate on the usefulness of this form of analysis. At one extreme is the view that relatively straightforward measures of a person’s language, such as the proportion of two and three letter words, produced in any circumstance or context, whether writing or speaking, can be utilised to determine whether utterances are those of one or more people. At the other extreme is the view that only with very large amounts of material uttered under very similar circumstances can distinctions between people be determined objectively and these are likely to be particular idiosyncrasies that differ from person to person rather than general aspects of verbalisation. Professor Canter placed himself firmly at the latter end of the continuum. In his opinion there is no evidence that there are specific aspects of a person’s use of words that can be identified as being so specific to that person that they can be used to identify the utterances of that person under a variety of conditions.
In the course of this judgment it is quite impossible to set out the entire detail of Professor Canter’s opinion. His central challenge to CUSUM is to its fundamental claim that the divergence of the two lines on a CUSUM chart is an indicator of mixed authorship. To test such a claim, randomly selected material of single and mixed authorship would need to be subjected to CUSUM analysis without the analyst knowing which was single and which was mixed. The CUSUM analyst would then need to demonstrate that their procedure could reliably distinguish between single and mixed authorship. In any conventional scientific study the researchers would report the degree of effectiveness of their procedure.
Professor Canter further states that CUSUM technology has not found its place in the scientific literature. It is not accepted in courts in the United States of America following recent legal rulings that emphasise that any claims of scientific evidence must overcome the “presumption of unreliability” (see O’Connor & Krauss, 2001) which states that:-
The technique has no such level of acceptance. Professor Canter states that all independent researchers who have carried out studies have found that the CUSUM procedure does not give results that differ from those which would be obtained by mere chance. No objective criteria had been given for determining when the divergence appliance in a CUSUM graph is significant. The recent addition by Dr. Farringdon of “significance testing” to his use of CUSUM charts does not overcome the fundamental problem that he has to show that the statistical difference he finds has some substantive meaning in relation to authorship. A study by Canter & Chester (‘Investigation into the claim of weighted CUSUM in authorship attribution studies’ (1997) Forensic Linguistics 4 (2) 253-261) demonstrated that the statistical procedures used by Farringdon under the general heading of “Weighted CUSUM” does not improve the validity of this procedure above chance level. It merely indicates if the lines on the chart themselves have some notional difference from chance variability, not whether that difference has any relevance to determining authorship.
“A principle or method is presumed to be not reasonably reliable if it does not have substantial acceptance within the relevant scientific, technical or specialised community.”
He was also extremely critical of the many arbitrary qualities of CUSUM whereby punctuation and other editing is introduced by Farringdon to modify the text from what was actually said. The creation and interpretation of the CUSUM chart was thus arbitrary and subjective.
Professor Canter then gave a list of examples illustrating how claims for CUSUM had been undermined, including an exercise whereby Channel 4 in its documentary television programme “Street Legal” managed to prevail upon Rev Morton, the creator of CUSUM, to conduct a blind trial of a number of transcripts. When this was screened in June, 1993 it showed that Morton claimed from his CUSUM analysis that the material he was sent was the work of one person. In fact it was the text from two different authors.
During the course of his evidence, Dr. Farringdon first indicated that he knew very little about this particular episode, but upon being asked by the Court to furnish to it a document from which he was reading, it transpired that the document in question contained a detailed treatment of the Channel 4 programme in question.
Numerous other examples of failed CUSUM analysis appear at para. 18 of Professor Canter’s report. He notes that most of these case examples were reported in 1992 and 1993. He says that since that time there have been no further studies that provide any more support for CUSUM. He believes this indicates that the scientific community decided in the early 1990’s that CUSUM was a non-productive area of research.
In conclusion, Professor Canter does not suggest that Dr. Farringdon or his colleagues are in any way fraudulent, but rather that they cherish a mistaken belief in CUSUM because they do not in fact apply an appropriate scientific model to their studies.
At this stage of the proceedings, as counsel agreed, the function of this Court is simply to determine whether the newly discovered facts, if established as such in evidence, render the conviction unsafe and unsatisfactory so that the conviction in this particular case should be quashed.
As was noted by Lynch J. in The People (Director of Public Prosecutions) v. Pringle (No.2)  2 IR225 (at p. 240):-
At the following page (p. 241) Lynch J. continued:-
“A person who has been convicted of an offence on indictment and who has unsuccessfully appealed to the Court of Criminal Appeal has merely to allege subsequently in a further notice of application to that court that a new or newly discovered fact shows that there has been a miscarriage of justice whereupon he comes within the terms of s. 2 of the Act of 1993 and becomes entitled to apply for an order quashing the conviction........ all that the applicant had to do was to lodge a notice of application to the Court of Criminal Appeal alleging that a new or newly-discovered fact shows that there has been a miscarriage of justice in his case and he thereby automatically fulfilled the conditions required by s. 2 to bring his case for review by the Court of Criminal Appeal. By virtue of sub-s. 2 of s. 2, the application is to be treated for all purposes as an appeal to the court against the convictions of November, 1980. It follows that at that stage of the application there is no requirement to show an actual miscarriage of justice. Any defect or error in the trial such as would render the convictions unsafe and unsatisfactory is sufficient to lead to a quashing of the convictions and this of course reflects the first basic aim of the Act of 1993 as set out in the long title.
Once the simple mechanics which I have described above have been complied with the application becomes by virtue of sub-s. 2 of s. 2 an appeal to the Court of Criminal Appeal against conviction and is thereafter governed by s. 3 of the Act. Under sub-s. 1 of s. 3 the Court of Criminal Appeal may (a) affirm the conviction, (b) quash the conviction, (c) quash the conviction and order a retrial and (d) quash the conviction and substitute a conviction of another offence.”
By reference to what criteria should the court now assess and evaluate the evidence which it has heard in the context of this appeal?
“The mere fact that the Court of Criminal Appeal quashes a conviction and does not order a re-trial does not of itself establish that a miscarriage of justice has actually occurred.”
This question has already been determined by the Supreme Court in The People (Director of Public Prosecutions) v. Joseph Gannon  1 I.R. 40. The judgment of Blayney J. delivered in that case makes it clear that the test is not to enquire whether the new material rendered the conviction of the appellant unsafe and unsatisfactory having regard to the course actually taken by the defence at trial but rather to ascertain whether the defence could have used the material in such a way as to raise a doubt about a significant element in the prosecution case and the possibility that a different approach by the defence may have led to an acquittal. In the course of his judgment, Blayney J. stated as follows (at pp.47 to 48):-
Thus what the court is required to do is to carry out an objective evaluation of the newly discovered fact with a view to determining in the light of it whether the applicant’s conviction was unsafe and unsatisfactory in the context of what the legal advisors might have done with the material if it had been available to them. The court can not simply have regard only to the course actually taken by the defence at trial.
“The court could not conclude for certain that the advent of the newly discovered material would have no effect on the manner in which the defence was conducted. The furthest one could go would be to say that it is possible that it might not have had any effect and this would not relieve the court from examining what the position would have been if the defence had availed of the newly discovered material and altered its strategy accordingly”.
The Court must therefore evaluate the evidence in relation to the newly discovered fact, firstly to ascertain if the evidence establishes the newly discovered fact, secondly to assess the weight and credibility of such evidence by reference to an objective standard to determine whether defence counsel could have utilised the newly discovered material in such a way as to raise a reasonable doubt in the minds of the jury about a significant element in the prosecution case.
Turning, firstly, to the medical evidence, the Court is satisfied that the missing photographs do constitute evidence which warrant the description of being “newly discovered facts”. They were elements of the overall post mortem photographic evidence which were not given to the defence at the time of trial, although it is not suggested that this was done deliberately or that any suspicious circumstances surround their non-production at that time.
That said, the Court is satisfied that they would have been of little benefit to the defence in this case. Firstly, the Court has considerable doubts as to whether photographs of such a prejudicial nature would ever have been introduced in evidence and believes that the defence would have been more likely to object to their proof than anything else. Secondly, having considered the expert evidence of Dr. Purdue and Professor Crane, and while of the view that these internal photographs – taken in isolation – would not have confirmed bruising of the neck or a diagnosis of cause of death as either strangulation or neck compression, the Court must nonetheless bear in mind that these photographs are not to be taken in isolation on the issue of causation, but must be considered in the context of the other photographs, the position of the body as subsequently found, the examination, findings and evidence of Dr. Harbison and, naturally, the evidence of the two witnesses heard on the appeal.
Taking all this evidence together, the Court is greatly assisted by the evidence given by Professor Crane. The Court specifically notes that Professor Crane based his conclusions not only on the findings made by Dr. Harbison, but also on what he saw as findings of considerable significance, namely the presence of areas of discolouration on the right parietal area of the skull of the deceased and over her right jaw. This patchy area of discolouration was on the side of the body which was uppermost in the bed and thus, in Professor Crane’s opinion, was not what one would expect to find if discolouration had been caused by putrefaction of staining. One would expect uniform discolouration in either of those circumstances. Bloodstaining tends to follow the direction of gravity and was more likely in this instance to be found on those portions of the body in contact with the bed. The body was burnt predominantly on the left side where it was in contact with the bed. Putrefaction tends to be uniform where it is evident. The Court is satisfied there is thus a very sound basis for the opinion of Professor Crane that, whatever about the findings in relation to the neck, the findings in relation to the skull and jaw are bruising and nothing else.
In relation to the neck it is clear that bruising cannot be excluded. That it was bruising was the professional opinion of Dr. Harbison whose many, many years of professional expertise must be accorded proper and due weight in any review of the type being undertaken in this case. He was the man on the spot and the person best able to make an assessment from the primary material consisting of the remains of the deceased. No witness can exclude compression of the neck as a cause of death. Ultimately the difference between this diagnosis and one of strangulation is more a matter of semantics than substance. What must be excluded on the evidence, however, is any accidental or natural cause of death. Indeed, the Court considers the possibility of both Martin Glynn and Margaret Glynn dying of natural causes within minutes or a few hours of each other as something which is totally ruled out on the evidence in this case, particularly having regard to the occurrence of the fire which clearly took place after Margaret Glynn had died. The Court does not believe therefore that any further cross – examination of Dr. Harbison by reference to the absent photographs would have thrown any real doubt on his diagnosis of cause of death
In relation to the CUSUM analysis, the Court in this case prefers the evidence of Professor Canter in relation to the reliability of the CUSUM technique. The Court is not satisfied that the technique has a properly established scientific provenance or that it has achieved the requisite degree of expert peer approval. The Court accepts the various criticisms made of the technique by Professor Canter, but will refrain from making any declaration which would have the effect of treating the CUSUM technique as bereft of scientific value. The Court will simply confine itself to saying that the onus of proof has not been discharged in this case to satisfy the Court that it is a science upon which reliance can be placed in this particular case. It may well be that further scientific advances in this area may yield a quite different result and outcome in other cases in future years.
The Court was also left in a state of considerable unease by the evidence given by Dr. Farringdon. On several occasions he declined to answer questions put to him in cross-examination, or prevaricated before giving a reply which was often not a reply to the particular question. In particular, the court noted that Dr. Farringdon, when invited to do so, was unwilling to submit to a blind trial in respect of the CUSUM technique, or at least offered various reasons why he should not do so. The Court acknowledges, however, that Dr. Farringdon is now in retirement and that the imposition of such a requirement might have been a particularly severe burden in the circumstances
It follows therefore that the Court does not accept that a “newly discovered fact” has been established under this heading of appeal and, that being so, it is unnecessary to say anything more. It follows that the Court is satisfied that the confession of the applicant, which dovetails with the other evidence led by the prosecution in this particular case, was given in a voluntary manner and was given by the applicant alone.
The applicant has not, in respect of the one newly discovered fact established in evidence, put in question the safety of the conviction in this case. The court will therefore refuse the application made under s. 2 of the Criminal Procedure Act, 1993.