Presenting physics: talks, posters and the oral defence

Physics is communicated in talks and posters as much as in papers. A conference talk may be the first time anyone outside your group hears about your work; a poster session is where many collaborations begin; and the oral defence of an honours thesis is where examiners decide how well you understand what you did. These forms reward different skills from writing. A reader can reread a paragraph; a listener cannot rewind a sentence. A reader chooses the pace; a listener follows yours. This lesson covers how to plan and deliver a short research talk, how to design a poster, and how to prepare for the questions in a defence, with Maya's twelve-minute departmental talk, her conference poster and her oral examination as examples.

Planning a research talk

Most student research talks are short: ten to fifteen minutes plus questions at an undergraduate conference or a departmental honours session. That is far too little time to present everything in the thesis, and trying to do so is the commonest mistake. A short talk can carry one main message, supported by perhaps three key results. Decide what that message is before making any slides. Maya's was: a parameter-free model that treats the magnet's finite size predicts its fall through a pipe to within about 3 per 100, while the textbook point-dipole formula fails by up to 37 per 100 in a way that depends on the gap between the magnet and the wall. Everything in the talk served that sentence.

A good structure for a short talk follows the shape of the thesis but with different proportions. Spend about a fifth of the time on motivation and the question, so the audience knows why to care; about a quarter on the method, including only what the audience needs to trust the results; about two fifths on the results; and the rest on conclusions and what comes next. Show the question early and the answer clearly, and repeat the main message at the end. Audiences remember beginnings and endings; the middle is where they need help, in the form of clear transitions that say where the talk is going.

A common guideline is about one slide per minute, but the right number depends on the slides. A slide with one plot and a single message may need ninety seconds; a title slide needs ten. What matters is that each slide has a clear purpose, which is best stated in its title as a sentence rather than a label. A slide titled "Results" tells the audience nothing; one titled "The point-dipole model fails by an amount that grows as the gap shrinks" tells them what to look for in the plot below it. Put as few words on slides as possible, and never read them aloud; the audience reads faster than you speak and will stop listening. Make plots larger and simpler than in the thesis: thicker lines, larger labels, fewer curves, and an explanation of the axes every time a new plot appears.

slides = [   # (title, seconds planned)
    ("Title: eddy-current drag on a falling magnet", 20),
    ("A magnet falls slowly through a copper pipe", 50),
    ("Why it matters: models of magnetic braking", 60),
    ("Point dipole versus a finite magnet", 70),
    ("Apparatus: light gates and five pipes", 60),
    ("Calibrating the moment: a 1.5 mm surprise", 60),
    ("Results I: point dipole fails by 17 to 37 per 100", 80),
    ("Results II: the failure follows the gap", 60),
    ("Results III: finite magnet agrees to 2.7 per 100", 70),
    ("Results IV: a cold pipe slows the magnet", 50),
    ("Conclusions and next steps", 60),
]
limit = 12*60
total = sum(s for _, s in slides)
print(f"{len(slides)} slides, {total//60} min {total%60:02d} s planned, limit {limit//60} min")
t = 0
for title, s in slides:
    t += s
    flag = "  <- main result" if "agrees" in title else ""
    print(f"  {t//60:2d}:{t%60:02d}  {title}{flag}")
words_per_min = 140
print(f"script length for a {words_per_min} words per minute pace: about {total/60*words_per_min:.0f} words")

11 slides, 10 min 40 s planned, limit 12 min
   0:20  Title: eddy-current drag on a falling magnet
   1:10  A magnet falls slowly through a copper pipe
   2:10  Why it matters: models of magnetic braking
   3:20  Point dipole versus a finite magnet
   4:20  Apparatus: light gates and five pipes
   5:20  Calibrating the moment: a 1.5 mm surprise
   6:40  Results I: point dipole fails by 17 to 37 per 100
   7:40  Results II: the failure follows the gap
   8:50  Results III: finite magnet agrees to 2.7 per 100  <- main result
   9:40  Results IV: a cold pipe slows the magnet
  10:40  Conclusions and next steps
script length for a 140 words per minute pace: about 1493 words

Delivering the talk

Rehearse aloud, standing, with a timer, at least three times, and at least once in front of people who will tell you honestly what was unclear. The first rehearsal is always too long; cut material rather than speaking faster. Know the first two minutes nearly by heart, since nerves are worst at the start, and know how you will finish. Speak to the audience, not to the screen; point at the feature on a plot that you are talking about; and pause after showing a key result so that people have time to look at it. If you run short of time, skip a slide deliberately rather than rushing through all of them. Most conference sessions have a chair who will stop you at the time limit, and the result you did not reach is the one nobody hears.

Questions after a talk are an opportunity, not an ambush. Listen to the whole question, repeat or rephrase it so the audience hears it, and answer the question that was asked, briefly. If you do not know, say so and say how you would find out: "I haven't tested that, but the code could estimate it by changing the magnet's length". If a question reveals a genuine problem, acknowledge it. Preparing a few backup slides for likely questions, such as the full uncertainty budget or the derivation of a key formula, lets you answer precisely without cluttering the main talk.

Designing a poster

A poster is read, or more often glanced at, by people walking past in a crowded hall. It must attract attention from several metres away, communicate its main point in about thirty seconds, and support a five-minute conversation with someone who stops. That means a large, informative title that states the result, a layout with a clear reading order, usually in columns from top left to bottom right, large figures, and few words. A useful target is fewer than 800 words in total, with body text readable from two metres, which for a standard poster means at least 24 point. Put the main result where the eye falls first, near the top centre, and make sure a passer-by who reads only the title and the largest figure takes away the message.

sections = {   # section: words of text
    "title": 14, "question": 90, "models": 120, "apparatus and calibration": 110,
    "main result caption": 60, "residuals caption": 40, "cold pipe": 80,
    "conclusions": 70, "references and acknowledgements": 60,
}
words = sum(sections.values())
print(f"total words: {words} (target under 800)")
reading_time = words/200*60
print(f"time to read every word at 200 words per minute: {reading_time:.0f} s")
print("largest sections:", sorted(sections, key=sections.get, reverse=True)[:3])
view_m = 2.0
min_cap = view_m/0.33              # rough legibility rule: 1 mm of letter height per 0.33 m of distance
print(f"readable from {view_m} m: letter height at least about {min_cap:.0f} mm, roughly 24 point or larger")

total words: 644 (target under 800)
time to read every word at 200 words per minute: 193 s
largest sections: ['models', 'apparatus and calibration', 'question']
readable from 2.0 m: letter height at least about 6 mm, roughly 24 point or larger

At the session, stand beside the poster, not in front of it, and have a two-minute spoken summary ready for anyone who stops: the question, the method in a sentence, the main figure and the conclusion. Let the visitor guide the conversation after that; some will want the physics, some the apparatus, some your future plans. Bring printed copies of the abstract or a link to the thesis and data. Many of the most useful conversations in a research career begin at posters, and a visitor who asks a hard question is often the person most worth talking to.

The oral defence

The oral defence, or oral examination, usually follows the thesis talk. It typically lasts thirty to sixty minutes with the supervisor and one or two examiners, and its purpose is to establish that you understand your work, its foundations and its limitations, and that the thesis is your own. Examiners will have read the thesis and will have marked passages to ask about. Expect questions at several levels: about the basic physics behind your project, such as why a moving magnet induces currents and why they oppose the motion; about specific steps, such as how the uncertainty in the moment was estimated; about choices, such as why ten drops per pipe and not twenty; about limitations and alternatives, such as what would change with a rectangular magnet; and about the broader context, such as where eddy-current braking is used in practice.

Preparation is straightforward but must be thorough. Reread the thesis as an examiner would, marking every claim you would want to challenge, and prepare an answer for each. Make sure you can derive every key equation without notes and explain every figure. Know your uncertainty budget and which term dominates. List the weaknesses of your work honestly; examiners respect a student who identifies a limitation before they do. Ask your supervisor or classmates to hold a mock defence. During the examination, take a moment to think before answering, use the board for derivations, and say clearly when you are unsure. "I don't know, but here is how I would approach it" is a much better answer than a confident guess. Examiners are not trying to fail you; they are trying to find the edges of your understanding, and showing that you know where those edges are is a sign of mastery.