Conditional Agreements, Group Weapon Awareness and No-Case Submissions in Secondary Liability for Murder: Commentary on R v Bagge & Anor [2025] EWCA Crim 402

1. Introduction

The decision of the Court of Appeal (Criminal Division) in R v Bagge & Anor [2025] EWCA Crim 402 is an important post‑Jogee authority on secondary liability for murder in the context of group violence involving weapons. It addresses:

How a “conditional agreement” within an armed group can ground liability as a secondary party for murder;
The proper approach under Galbraith to a submission of no case to answer in a circumstantial case alleged to rest on such an agreement;
The limits of the “overwhelming supervening act” doctrine from Jogee where a principal uses a firearm;
The treatment of potentially prejudicial evidence of a co‑accused’s “bad character” and the protection of a defendant by judicial directions; and
The fairness and Article 6–type implications when a co‑defendant gives evidence‑in‑chief but then refuses to be cross‑examined.

The case arises from the murder of Tyler McDermott in Tottenham on 13 April 2023 and the subsequent convictions for murder of several young men, including the applicants, Kalam Bagge and Rhys Antwi. It is a carefully reasoned judgment by William Davis LJ (sitting with others) refusing leave to appeal, but in substance functioning as a full appeal judgment (¶¶3–4, 40).

Although the Court did not purport to create new law, it clarifies and consolidates several important principles on:

the evidential sufficiency threshold at half‑time in circumstantial secondary liability cases; and
the circumstances in which a jury may infer a murderous conditional agreement from group conduct and shared awareness of weapons, even in the absence of any overt assisting act by particular defendants.

There is also a procedural context: a reporting restriction under section 4(2) of the Contempt of Court Act 1981 postponed publication of the judgment because of a related forthcoming trial (¶1), illustrating the court’s ongoing role in balancing open justice with protection of pending proceedings.

2. Background and Parties

2.1 The Incident

In the early hours of 13 April 2023, a group of young men, including the applicants Bagge and Antwi and co‑accused Barnett and Reid, congregated at Pirate Studios in Tottenham where a party was taking place (¶9).

Key factual features, largely reconstructed from CCTV, included:

Prior to Bagge’s arrival, Antwi had been handed a large knife by a female in a side area off a corridor; he showed it to Barnett and another male, then made a gun gesture towards Barnett (¶10).
When Bagge arrived with Reid and a female, there was an interaction at the party room door where the female (carrying a plastic bag) and Barnett stood close together for long enough for an object to be transferred (¶9).
Shortly afterwards, Barnett, Bagge and Reid moved to a secluded area; Barnett appeared to show them something; on emerging, something blue (later linked to the blue plastic bag containing the gun) was visible in Barnett’s pocket (¶9).
There was visible tension in the corridor between the group and McDermott’s group; Reid at one point drew out the sword‑like knife (later used to stab McDermott) in a way that made it plainly visible (¶11).
At around 4 am, the group including Bagge and Antwi left together and stood as a group across the road from the venue exit (¶12).
When McDermott emerged and approached, Barnett very quickly produced the gun, shot McDermott in the head and then fired at one of McDermott’s friends (¶12).
As Barnett and others (including the applicants) ran off, Reid turned back, attacked McDermott’s prone body with the sword‑like knife causing a gaping wound, and then ran off to rejoin the group (¶12).
Subsequent CCTV showed Bagge at Walthamstow bus station dropping and retrieving what appeared to be a knife (¶13); Antwi later travelled to Northampton and stayed at a Travelodge (¶13).

The pathology evidence indicated that while the knife wound could have been fatal, the gunshot was treated as the substantial cause of death, likely causing immediate death (¶5).

2.2 The Prosecution Case

The prosecution’s core theory was the existence of a conditional agreement among Barnett, Reid, the applicants and others that, should the need arise, one or more of them would use unlawful violence against others, intending at least to cause really serious harm (¶6).

The Crown alleged that:

Barnett’s shooting and Reid’s stabbing were acts carried out in furtherance of that agreement (¶6); and
By entering into the agreement, Bagge and Antwi assisted or encouraged Barnett, the principal offender (¶6).

2.3 Convictions and Procedural Posture

Bagge and Antwi were convicted of murder on 19 June 2024 at the Central Criminal Court (¶2).
Barnett was convicted of murder and attempted murder; Reid of murder (¶2).
Two co‑accused were acquitted; for two others, no verdicts were returned and a retrial was pending, prompting the Registrar to refer the applications for leave directly to the full court because of the imminence of the further trial (¶3).

Bagge and Antwi sought leave to appeal against conviction. Antwi also advanced two discrete additional grounds:

Barnett’s absence from cross‑examination – arguing that his voluntary failure to return to court after giving evidence‑in‑chief rendered it unfair to continue the trial in Antwi’s case (¶8, ¶17); and
Admission of bad character evidence concerning Reid’s purchase of “Roman sword knives” in 2022, said to have prejudicial spill‑over onto Antwi (¶8, ¶15).

The Court approached the applications as if they were full appeals, with both sides represented by leading counsel (¶4).

3. Summary of the Judgment

The Court of Appeal refused leave to appeal, holding that none of the grounds advanced rendered the convictions unsafe (¶40).

In outline:

The evidence of Reid’s earlier knife purchases was properly admitted in his case and did not prejudice Antwi, given a clear judicial direction to disregard it in relation to him (¶¶20–22).
The trial judge was right not to discharge the jury in Antwi’s case when Barnett refused to attend for cross‑examination; Antwi’s ability to challenge Barnett’s evidence was not materially prejudiced (¶¶24–25).
The trial judge applied the correct Galbraith test in rejecting submissions of no case to answer by both applicants, and was entitled to leave their cases to the jury (¶¶27, 33, 38–39).
There was sufficient evidence on which a reasonable jury could infer a conditional agreement to use unlawful violence with intent to cause really serious harm, binding the applicants as secondary parties (¶¶30–31, 34, 38–39).
Barnett’s use of the gun was not an “overwhelming supervening act” breaking any causal or liability chain under Jogee (¶35–36).
The judge’s summing‑up on agreement and secondary liability was unobjectionable, even though there was no overt assisting act by the applicants; the agreement itself could constitute assistance or encouragement (¶38–39).

The core doctrinal significance lies in the Court’s treatment of:

how the no‑case test applies in circumstantial secondary liability cases; and
the evidential basis on which juries may infer murderous conditional agreements in armed group scenarios post‑Jogee.

4. Detailed Analysis

4.1 The Discrete Grounds in Antwi’s Case

4.1.1 Bad Character / “Roman Sword Knives” Evidence

During the prosecution case, the Crown sought to introduce evidence that in October and November 2022 Reid had an account with a Luton company selling “Roman sword knives” and had purchased eight such knives (¶15). These were very similar to the large knife found at Antwi’s home on 20 April 2023, itself similar to the knife handed to Antwi that night (¶10, ¶15).

The prosecution’s theory:

Reid had supplied the knife found at Antwi’s home; and
this was relevant to whether Reid knew that others in the group, including Antwi, were armed, which in turn was relevant to the existence of the alleged agreement (¶15).

The trial judge admitted the evidence:

as “evidence to do with the facts of the offence” (i.e. integral factual context rather than mere propensity evidence); and
in the alternative, under section 101(1)(d) Criminal Justice Act 2003 as evidence of bad character relevant to an important issue in the case (¶15).

Crucially, in summing‑up she directed the jury that this evidence was irrelevant to the case against Antwi (¶15, ¶21).

On appeal, Antwi argued:

the evidence was too remote in time (six months earlier) and invited improper speculation that Reid must have known Antwi was armed (¶18); and
its admission was prejudicial to Antwi, especially in light of the perceived relationship between him and Reid (¶¶18, 22).

Reid’s counsel also objected at trial that the purchases were too remote to have probative value (¶19).

The Court of Appeal held:

Admissibility as against Reid: The judge was entitled to find that the knives were “singular in appearance” and very similar to the knife in Antwi’s possession. This gave them some probative value on whether Reid supplied Antwi’s knife and knew he was armed (¶19–20). Even if not compelling, they were legitimately part of the evidential picture.
No prejudice to Antwi:
- The judge gave clear and unequivocal directions to ignore the evidence in relation to Antwi (¶21–22).
- Antwi’s own case accepted that he had a large knife that night (¶21).
- Even if he said he obtained it from Reid, the Court considered that would not materially affect his case (¶21).
- The suggested prejudice, that the jury might speculate about the nature of the relationship between Antwi and Reid, was neutralised by the judge’s direction; the Court reiterated the conventional stance that one must proceed on the basis that the jury followed the direction (¶22).

From a doctrinal perspective, two points stand out:

The Court endorses a relatively flexible, fact‑sensitive approach to treating prior acquisitions of distinctive weapons as “evidence to do with the facts of the offence” in group violence cases.
It reaffirms the centrality of limiting directions to protect co‑defendants from spill‑over prejudice, and the appellate court’s usual assumption that juries comply with such directions unless there is strong reason to think otherwise.

4.1.2 Barnett’s Absence and the Refusal to Discharge the Jury

After the prosecution case and the refusal of the no‑case submissions, Barnett gave evidence. He claimed:

he had found the gun outside an earlier party that night and given it to the female, who then returned it to him at Pirate Studios (¶16);
he said nothing about the gun gesture made by Antwi, which was plainly visible on CCTV (¶16).

Barnett subsequently refused to return to court for cross‑examination. After investigation, the judge concluded that:

his absence was voluntary and not due to illness; and
the prejudice to the female co‑accused who was alleged to have handed over the gun was irremediable, so the jury was discharged in her case (¶17).

However, the judge refused to discharge the jury in Antwi’s case. Antwi argued on appeal that:

he was deprived of the opportunity to cross‑examine Barnett about:
- the alleged gun gesture (to suggest it was innocuous);
- Barnett’s statement that he had heard McDermott’s group had weapons; and
- the credibility of Barnett’s explanation for how he acquired the gun (¶23–25).

The Court rejected this ground:

It held that none of the matters Antwi wished to put to Barnett would have advanced his case or that his position was prejudiced by the inability to cross‑examine (¶24).
As to undermining Barnett’s account of finding the gun in the street, the Court considered that this was obviously incredible without cross‑examination, and that even if counsel had demonstrated that, it would not have materially helped Antwi (¶25).
On the gun gesture and Barnett’s knowledge, the CCTV itself showed the gesture, and Barnett’s omission to mention it in chief arguably spoke for itself; there was little realistic forensic advantage to be gained by questioning him further.
At the very least, the judge’s decision not to discharge was “well within the range reasonably open to her” (¶25) – signalling the Court’s deference to trial judges’ evaluative case management in complex multi‑handed trials.

Although not couched explicitly in Article 6 ECHR terms, the decision is consistent with the Strasbourg and domestic jurisprudence: the right to examine witnesses is not absolute and must be assessed in terms of overall fairness and the materiality of the lost opportunity. The Court stresses substance (did it make the trial unfair?) rather than form (was there any theoretical cross‑examination that might have occurred?).

4.2 The Core Issue: No Case to Answer and Conditional Agreement

4.2.1 The Judge’s Ruling and the Galbraith Test

At the close of the prosecution case, both applicants submitted that there was no case to answer because the evidence was insufficient for a reasonable jury to conclude that:

the alleged conditional agreement existed; or
either applicant was a party to it (¶7, ¶26).

The trial judge:

identified the well‑known test in R v Galbraith [1981] 1 WLR 1039 (¶27); and
set out the approach in circumstantial cases, explicitly rejecting the proposition that the prosecution must, at half‑time, have excluded all reasonable possibilities consistent with innocence (¶27).

Her formulation (¶27) was:

“The question is whether there is evidence, in the case of any particular defendant, on which a jury, properly directed, could infer guilt. In deciding whether there is a case for any particular defendant to answer, it is not my function to decide whether I can reject all reasonable possibilities consistent with innocence; rather, the question is whether a reasonable jury could, on one possible view of the evidence, conclude that a defendant was guilty.”

The applicants argued on appeal that, particularly in light of Bassett [2020] EWCA Crim 1376, the judge was required herself to be satisfied that all reasonable possibilities consistent with innocence had been excluded, and erred in failing to engage in that exercise (¶32).

The Court of Appeal firmly rejected this reinterpretation of Galbraith:

It endorsed the judge’s formulation as correct: she asked whether a reasonable jury, properly directed, could exclude any reasonable possibilities consistent with innocence (¶33).
It quoted from R v Younis Masih [2015] EWCA Crim 477, ¶3 (¶33):

“Matters of assessment and weight of the evidence are for the jury and not for the judge. Since the judge is concerned with the sufficiency of evidence and not with the ultimate decision the question is not whether all juries or any particular jury or the judge would draw the inference of guilt from the evidence adduced but whether a reasonable jury could draw the inference of guilt.”

Thus, the Court clarifies a line of authority culminating in Bassett:

In simple, factually narrow cases (like Masih itself), it may be necessary for the judge to consider whether alternative innocent explanations are so compelling that no reasonable jury could properly convict (¶27).
But that does not shift the basic Galbraith test: the judge does not perform the jury’s function of actually weighing competing inferences and deciding whether to reject the innocent ones. The judge asks only if a reasonable jury could, on one view of the evidence, draw the inference of guilt (¶33).

This is a significant clarification for circumstantial cases generally, and particularly for secondary liability cases where the Crown relies on inferences from group conduct rather than direct evidence of an explicit agreement.

4.2.2 Secondary Liability Post‑Jogee and the Role of “Agreement”

The trial judge then considered the principles of secondary participation in light of R v Jogee [2016] UKSC 8 (¶28), which had corrected the law on joint enterprise. She distilled the prosecution’s burden for a secondary party:

Conduct element: assistance or encouragement of the principal’s offending;
Mental element: an intention to assist or encourage the principal to attack the victim with an intent to kill or cause really serious harm (¶28).

She referred to Jogee’s discussion of agreement as one evidential route to proving assistance/encouragement. At ¶78 of Jogee, quoted by the judge (¶29), the Supreme Court explained:

Secondary liability does not require an agreement;
But where an agreement exists, it is by its nature a form of encouragement and usually involves acts of assistance: agreeing to carry out a criminal venture is itself evidence of intention to assist;
People can also become secondaries without agreement, via spontaneous supportive presence, words, or deeds.

Against that background, the judge defined the alleged agreement (endorsed by the Court of Appeal) as a conditional agreement (¶30):

“The agreement alleged in this case is a conditional agreement: that ‘should the need arise, one or more of them would use unlawful violence upon another or others, intending to cause that other person or persons really serious injury.’ The defendants may have joined the agreement at different points in the course of the evening.”

This definition is important. It recognises that group violence agreements are often contingent: the group arms itself and stands ready to use serious violence if a perceived trigger occurs, rather than setting out with a fixed plan to attack a named person at a specific moment.

4.2.3 Evidence Supporting a Conditional Agreement

The judge identified, and the Court endorsed, a series of factual inferences a reasonable jury could draw (¶31):

All members of the group were aware that Reid and Antwi had large knives.
All members knew that Barnett had a gun:
- Bagge from having seen it, including in the secluded corridor episode where Barnett appeared to show something and later had something blue visible in his pocket matching the gun bag (¶9, ¶31);
- Antwi from the earlier gun gesture he made after showing Barnett his knife (¶10, ¶31).
The group spent a substantial period together in and around the corridor, in close discussion, while tensions with McDermott’s group were apparent (¶11, ¶31).
They left together and waited outside the venue as a group until McDermott appeared (¶12, ¶31).
After the shooting, they ran off together (¶12, ¶31).

From these features, the judge concluded that:

There was material for a reasonable jury to find an agreement existed; and
Given the level of weaponry and mutual awareness, it was a permissible inference that the agreement was to use unlawful violence if “the need arose” with intent at least to cause really serious injury (¶31, ¶34).

The Court agreed, describing as “equally valid” an analysis that:

the group was well armed; and
they waited as a group for McDermott and his associates to emerge, in circumstances where they knew what weapons were available within the group (¶34).

On this footing, the Court held that there was sufficient evidence from which a reasonable jury could find the existence of the conditional agreement and the applicants’ participation in it (¶¶34, 38–39).

4.2.4 Overwhelming Supervening Act: Was Barnett’s Use of the Gun a Break in Liability?

The applicants also suggested that even if there had been an agreement to use violence, Barnett’s decision to use the gun was an “overwhelming supervening act” as described in Jogee at ¶97, which would relieve them of liability (¶35).

Jogee recognises that:

“… it is possible for death to be caused by some overwhelming supervening act by the perpetrator which nobody in the defendant’s shoes could have contemplated might happen and is of such a character as to relegate his acts to history; in that case the defendant will bear no criminal responsibility for the death.”

The Court of Appeal noted:

The trial judge had not dealt explicitly with this issue in her half‑time ruling, likely because it had been argued that this issue should await the conclusion of all the evidence (¶36).
In any event, on the judge’s factual findings, there was evidence that the applicants did contemplate the use of a gun, by virtue of their awareness that Barnett had one (¶36).
Even if knowledge of the gun was wrongly inferred, there was still prima facie evidence that the applicants knew members of the group were carrying dangerous weapons and that they themselves were so armed (¶36).
On that basis, the use of a gun was an escalation of the level of weaponry, not something so extraordinary as to relegate the prior agreement to history (¶36).

Thus the doctrine of overwhelming supervening act was inapplicable. The Court makes a subtle but important point:

Where a group has a conditional agreement to use serious violence, and the principal uses a more dangerous weapon than others anticipated, this may in some cases be merely an escalation rather than a historical break.
Only where the principal’s act is truly unforeseeable in kind – not merely in degree – and of such character that earlier participation is rendered irrelevant, will the secondary be exonerated.

4.2.5 No Overt Act and the Role of the Agreement Itself

A central plank of the applicants’ submissions was the absence of any overt assisting act by either Bagge or Antwi at the scene:

Neither produced a knife during the incident (only Reid did);
They were merely present; and
The violence outside the venue was said to be “spontaneous” and unconnected with any pre‑existing agreement (¶37).

Counsel for Antwi (Mr Green KC) further argued that convicting the applicants of murder on the basis of an agreement without an overt act “sent out a poor message” to young men who carry weapons for self‑defence (¶37).

The Court responded in two key ways:

The court’s function is not to send messages:
- William Davis LJ stated: “With respect to Mr Green, our function is not to send out messages. Rather, it is for us to determine whether the evidence at the close of the prosecution case was capable in law of establishing secondary liability.” (¶38).
- This emphasises the judicial commitment to principled adjudication over policy‑messaging, at least in the context of the appellate function.
The agreement can itself be the assistance/encouragement:
- The Court highlighted the rationale of Jogee at ¶78 (agreement inherently constitutes encouragement and often assistance) and applied it to the facts.
- It accepted there was no evidence of overt assistance by the applicants in Barnett’s attack at the moment of shooting (¶38).
- However, the prosecution case was that there was a prior agreement that Barnett should use deadly force if the need arose; the Court held that “The agreement of itself amounted to assistance or encouragement” (¶38–39).

This is a critical doctrinal clarification:

Secondary liability does not require that the defendant physically assist or verbally encourage at the moment of the principal offence.
Where a conditional agreement to use serious violence with the necessary intent is proved, the making and maintenance of that agreement can itself satisfy the conduct element, particularly within an armed group moving as a unit towards a confrontation.

The Court also rejected criticism of the judge’s summing‑up, where she did not explicitly mention that “in most cases” there will be an overt act (a phrase drawn from Jogee ¶78) (¶39). It held:

The judge properly focused on the facts of the case rather than drawing comparisons with other cases (¶39).
She correctly directed that there was no overt act and left it to the jury to decide whether, notwithstanding that, the conditional agreement existed and encompassed Barnett’s use of deadly force (¶39).

4.3 Precedents and Authorities: How They Were Applied

4.3.1 R v Jogee [2016] UKSC 8

Jogee is the foundational modern authority on secondary liability and joint enterprise. The Court in Bagge draws on two main aspects:

Nature of Secondary Liability:
- Liability rests on proof that the alleged secondary intentionally assisted or encouraged the principal and intended that the principal would commit the crime with the necessary mens rea.
- Mere foresight that the principal might commit the crime is insufficient, though foresight can be evidence of intent.
Forms of Participation (¶78 of Jogee, quoted at ¶29):
- Agreement is one route to secondary liability: where parties agree to a criminal venture, each is liable for acts to which they have expressly or impliedly assented, because the agreement itself embodies encouragement and often assistance.
- But secondary liability is not limited to agreements; it also includes spontaneous supportive presence or other encouragement without any prior plan.
- The Court in Bagge uses this to legitimate the “conditional agreement” analysis and to emphasise that a secondary can be liable based on an agreement alone.
Overwhelming Supervening Act (¶97 of Jogee, discussed at ¶35–36):
- A secondary may escape liability if the principal commits an act so unexpected and extreme that no one in the secondary’s position could have contemplated it, thereby relegating the secondary’s prior contribution to history.
- The Court in Bagge holds that Barnett’s use of the gun, in the context of a group already armed with dangerous weapons and allegedly aware of the gun, did not meet that threshold.

Bagge thus confirms that Jogee’s more defendant‑friendly correction of the law does not insulate those who knowingly arm themselves and remain part of an armed group with a shared conditional intent to use serious violence.

4.3.2 R v Galbraith and R v Younis Masih

Galbraith sets the basic no‑case test: a case should be withdrawn from the jury where the prosecution evidence, taken at its highest, is such that a properly directed jury could not reasonably convict.

Masih (quoted at ¶33) re‑emphasises:

The judge’s role at half‑time is limited to sufficiency of evidence; matters of assessment and weight are for the jury.
The question is whether a reasonable jury could draw the inference of guilt, not whether they necessarily would, nor whether the judge personally would.

Bagge builds on Masih and responds implicitly to more recent cases like Bassett, in which defendants had argued that in circumstantial cases judges must themselves weed out alternative innocent explanations. The Court accepts that in some very simple cases the judge may properly do so, but makes clear that the fundamental test remains Galbraith‑based.

4.3.3 Bassett [2020] EWCA Crim 1376

Although the judgment does not quote Bassett, it is presented as the culmination of a line of authority relied upon by the applicants to argue for a more stringent half‑time scrutiny (¶32).

The Court essentially limits any broad reading of Bassett:

Judges remain bound by the orthodox Galbraith/Masih approach.
Any suggestion that they must comprehensively exclude all reasonable innocent explanations at half‑time is rejected.
Instead, the focus is on whether there is any possible view of the evidence on which a reasonable jury could be sure of guilt.

4.4 Impact and Future Significance

4.4.1 For Secondary Liability in Group Violence Cases

Bagge will be particularly important in cases involving:

groups of young men associated with gangs;
multiple weapons (knives, firearms) within the group; and
ambiguous CCTV‑based reconstructions of group conduct before an outbreak of violence.

Key messages:

A conditional agreement to use violence “should the need arise” can sustain murder liability where:
- group members are visibly armed;
- others in the group know of this;
- they move and remain together in circumstances of visible tension; and
- violence follows swiftly when a perceived trigger (e.g. the appearance or approach of the rival group) occurs.
Even without an overt act by a particular defendant at the moment of the fatal attack, their prior participation in such an agreement can supply both:
- the conduct element (assistance/encouragement); and
- the mental element (shared intention to cause really serious harm).

For prosecutors, the case underscores the importance of assembling a careful narrative of group conduct (use of CCTV, movement patterns, weapon visibility, clustering, timing) to invite inferences of conditional agreement. For the defence, it illustrates the need to:

attack alleged knowledge of weapons and the inferences drawn from “supportive presence”;
place emphasis on alternative explanations (e.g. weapons carried for self‑defence) and spontaneity; and
frame the Jogee intent requirement (did this defendant really intend serious harm to be inflicted?).

4.4.2 On Submissions of No Case to Answer

The clarification of the Galbraith test in circumstantial cases is likely to be cited regularly:

Trial judges need not, and should not, attempt to “try the case at half‑time” by deciding whether all innocent explanations can be personally excluded.
Instead, they must ask: is there enough evidence that a reasonable jury could draw the inference of guilt, on one plausible view of the evidence, if they choose to reject the innocent explanations?

This will tend to:

limit the success of no‑case submissions in complex, evidence‑rich cases where multiple inferences are open; and
push arguments about alternative factual analyses (e.g. spontaneous violence, self‑defence motives for carrying knives) firmly into the jury’s domain.

4.4.3 On Overwhelming Supervening Acts

Bagge narrows the potential for defendants in group cases to rely on the Jogee “overwhelming supervening act” doctrine where a firearm is deployed:

If the group is already armed with serious weapons, and there is evidence that a gun is present or that the use of serious violence is contemplated, the use of a gun will not normally be regarded as so unforeseeable as to break liability.
The doctrine will remain reserved for truly exceptional and qualitatively different acts by the principal.

4.4.4 On Bad Character Evidence and Co‑Defendant Prejudice

The decision reinforces:

The scope for treating prior acquisition of distinctive weapons as “evidence to do with the facts” rather than mere propensity evidence, particularly where the weapon type recurs;
The reliance appellate courts place on limiting directions to manage cross‑contamination risks, and the difficulty defendants face in showing actual prejudice where such directions are clear and the defendant’s own case admits the underlying fact (here, possession of a large knife).

4.4.5 On Absent Witnesses and Co‑Defendants

Finally, the case provides a useful illustration of:

the defendant‑specific, fact‑sensitive analysis required when a witness (especially a co‑accused) is unavailable for cross‑examination;
the court’s willingness to discharge the jury for one co‑accused (the female alleged to have provided the gun) while continuing with others, where the prejudicial effect differs (¶17); and
the appellate court’s deference to trial judges’ assessments whether the lost opportunity to cross‑examine could realistically have made a difference to the fairness of the trial.

5. Complex Concepts Simplified

5.1 Secondary Liability and Joint Enterprise

Secondary liability arises where a person (the secondary or “accessory”) is held criminally responsible for a crime committed by someone else (the principal), because the secondary:

Assisted or encouraged the principal to commit the crime; and
Intended that their assistance or encouragement would help the principal to commit that crime (with the required intent).

The older “joint enterprise” doctrine sometimes blurred this by treating mere foresight that another might commit a crime as enough. Jogee corrected that: foresight is evidence, not the test. What matters is actual intention to assist or encourage the crime charged.

5.2 Conditional Agreements

A conditional agreement is an agreement to do something if a certain situation arises.

In group violence:

The group may not have a firm plan to attack anyone, but may arm themselves and implicitly or explicitly agree that if the rival group appears, or if there is some provocation, they will use violence.
If that agreement includes at least an intention to cause really serious harm, and someone is then killed in the course of the violence, members to the agreement can be liable for murder as secondary parties.

5.3 Overwhelming Supervening Act

This is a narrow exception where a secondary is not liable despite prior assistance or encouragement. It applies where:

The principal commits a further act that is:
- so extreme and unusual that nobody in the secondary’s shoes could reasonably have contemplated it; and
- of such a character that it makes the secondary’s earlier acts historically irrelevant to the death.

Example (simplified): A and B agree to commit low‑level robbery with mild force. B, unknown to A, secretly brings petrol and deliberately burns the victim to death. If this is found to be an overwhelming supervening act, A would not be liable for murder.

In Bagge, the Court held that using a gun, where the group is already armed and allegedly aware of the gun, is not so extraordinary as to count as an overwhelming supervening act.

5.4 “No Case to Answer” and the Galbraith Test

At the end of the prosecution case, the defence can argue there is “no case to answer”, meaning the judge should stop the case and direct acquittals because the evidence is too weak.

Under Galbraith:

If the prosecution evidence, even taken at its highest, could not lead a reasonable jury properly directed to convict, the judge must stop the case.
But where reasonable juries could legitimately differ on what inferences to draw, the case must go to the jury. The judge does not decide which inference is right; that is the jury’s job.

5.5 Bad Character Evidence and Section 101(1)(d) CJA 2003

“Bad character evidence” is broadly evidence about a defendant’s misconduct other than the conduct charged. It can be very prejudicial; the jury might think “if he did something bad before, he must be guilty now”.

Section 101(1)(d) of the Criminal Justice Act 2003 allows such evidence where it is relevant to an important issue between the defendant and the prosecution (for example, whether the defendant is telling the truth, whether they had a particular propensity, or knowledge in issue).

However:

Sometimes evidence that might look like bad character is really just part of the story of the offence (e.g. earlier possession of the same distinctive weapon used in the offence). Then it may be admissible as “evidence to do with the facts”, not as bad character.
Where evidence relates to a co‑defendant but risks prejudicing another, the judge can – and often must – give the jury a clear direction that the evidence is irrelevant to that other defendant and must not be used against them.

6. Conclusion: The Significance of Bagge & Anor

R v Bagge & Anor [2025] EWCA Crim 402 is an important clarification and reaffirmation of core principles in modern English criminal law on secondary liability and the handling of circumstantial evidence.

Its key contributions can be summarised as follows:

It confirms that, under Galbraith, the trial judge’s role at half‑time is limited to assessing whether there is evidence on which a reasonable jury could convict. The judge is not required personally to exclude all reasonable innocent explanations – that remains a matter for the jury.
It endorses the concept of a conditional agreement as a proper evidential route to establishing secondary liability post‑Jogee. In armed group situations, such an agreement can itself constitute assistance or encouragement with the requisite intent, even in the absence of any overt helping act at the scene by particular defendants.
It narrows the scope of the overwhelming supervening act doctrine, making clear that the deployment of a gun in a context where weapons are already present and known is typically an escalation, not a break in the chain of liability.
It reinforces the central role of judicial directions in managing potentially prejudicial evidence (such as a co‑accused’s prior weapon purchases) and demonstrates the appellate court’s assumption that juries will follow clear directions.
It illustrates a careful, defendant‑specific approach to fairness when a key witness (here, a co‑accused) refuses cross‑examination: the court may discharge the jury in one defendant’s case but continue in another’s, depending on where real prejudice lies.

Taken together, these points mean that individuals who:

join armed groups;
know of the presence of deadly weapons within the group; and
participate in the group’s movement and stance in the face of tension with others,

may properly face murder liability where a jury is sure that they shared a conditional agreement that serious violence would be used if “the need arose”. Simply refraining from brandishing or using their own weapon at the critical moment will not necessarily shield them, if the agreement itself evidences their intentional assistance or encouragement of the fatal attack.

In that sense, Bagge consolidates the post‑Jogee landscape: it rejects over‑expansive joint enterprise based on mere foresight, but equally rejects attempts to narrow secondary liability so far that those who knowingly contribute to armed group dynamics can escape responsibility when lethal violence follows the very risk their agreement created.

Conditional Agreements, Group Weapon Awareness and No-Case Submissions: Commentary on R v Bagge & Anor [2025] EWCA Crim 402