Riemann Console dot org

An open research record on the Riemann Hypothesis.

NAV READY · TYPE SHORTCUT · ENTER EXECUTES

The lanes of the primes

Series
Explain
Summary
A beginner-first journey from visible Prime Spring lanes through modular arithmetic and arithmetic progressions to Dirichlet's theorem, Dirichlet characters, L-functions and the Generalized Riemann Hypothesis.
Math Level
GENERAL
Index Excerpt
Dirichlet's theorem; primes in arithmetic progressions; arithmetic progressions; residue classes; Dirichlet characters; Dirichlet L-functions; primes modulo q; Generalized Riemann Hypothesis; Prime Spring

Prime numbers seem to arrive irregularly.

2, 3, 5, 7, 11, 13, 17, 19, 23, 29,…2,\ 3,\ 5,\ 7,\ 11,\ 13,\ 17,\ 19,\ 23,\ 29,\ldots

Sometimes two sit close together. Sometimes a larger gap opens between them. As we travel farther along the number line, primes become less common, but they never disappear.

That straight number line is the usual place to meet them.

But what happens if we bend it?

Suppose we take the positive integers in their ordinary order and wrap them around a widening spring. Then suppose we mark the primes and dim everything else.

Something striking happens.

Points that looked irregular on a line begin to gather into lanes, ribs and repeating empty spaces.

Some of those patterns are consequences of very simple divisibility.

Some lead into much deeper mathematics.

And one of the most natural questions raised by the picture cannot be answered by any picture at all.

It leads to Peter Gustav Lejeune Dirichlet, to a landmark theorem about primes, and eventually to a whole family of functions extending the mathematical world around the Riemann zeta function.

But we can start much closer to shore.

We can start by turning the spring.

When a number line becomes a spring

The Prime Spring takes the ordinary positive integers and places them in sequence along a helix.

The integers themselves do not change.

A prime remains prime. A composite remains composite. Consecutive numbers still occur in the same order.

Only their positions in space have changed.

Now choose six integers for each complete turn of the spring.

Number 1 occupies some angular position. Six numbers later, 7 comes back to the same angle. Six after that, 13 does the same.

So

1, 7, 13, 19, 25,…1,\ 7,\ 13,\ 19,\ 25,\ldots

line up.

So do

2, 8, 14, 20, 26,…2,\ 8,\ 14,\ 20,\ 26,\ldots

and

5, 11, 17, 23, 29,…5,\ 11,\ 17,\ 23,\ 29,\ldots

Viewed along the spring, these repeating positions appear as lanes.

Nothing mysterious has happened yet. We told the spring to repeat after six integers, so numbers six apart line up.

But now there is an interesting experiment available to us:

put the primes into those lanes and see which lanes they can occupy.

(Changing the spring changes the representation, not the arithmetic. This distinction will remain important throughout the article. A visual pattern may expose genuine number-theoretic structure, but the geometry itself does not create new prime numbers.)

Six ways to leave a remainder

Why exactly do those numbers line up?

Take 17 and divide it by 6\.

Two complete sixes fit inside 17, with 5 left over.

Do the same with 23\.

Three complete sixes fit inside 23, again with 5 left over.

And with 29:

four complete sixes, again with 5 left over.

So the numbers

5, 11, 17, 23, 29,…5,\ 11,\ 17,\ 23,\ 29,\ldots

share something very elementary.

When divided by 6, they all leave remainder 5\.

The numbers

1, 7, 13, 19, 25,…1,\ 7,\ 13,\ 19,\ 25,\ldots

all leave remainder 1\.

Because division by 6 can leave only the remainders

0, 1, 2, 3, 4, 5,0,\ 1,\ 2,\ 3,\ 4,\ 5,

every integer belongs to one of six repeating families.

This is the basic idea of modular arithmetic.

Mathematicians describe 17 and 23, for example, by writing

17≡23(mod6),17\equiv23\pmod6,

because they occupy the same remainder position when division is by 6\.

And we can write

17≡5(mod6)17\equiv5\pmod6

to say that 17 leaves remainder 5\.

Each complete family of integers sharing one of these positions is called a residue class.

The visual lane on our six-turn spring is therefore the geometric image of a residue class modulo 6\.

(If the notation is less intuitive than the example, keep hold of the example. “Modulo 6” means that we are organising the integers according to where they land in a repeating cycle of six. The glossary entries on modular arithmetic and residue classes unpack this more slowly.)

Now we are ready to ask what primality does to those six families.

Put the primes back in

A prime number greater than 2 cannot be even.

That immediately rules out three of our six residue classes:

0, 2, 4(mod6).0,\ 2,\ 4\pmod6.

Every number in those lanes is even.

The class 3 has a different problem:

3, 9, 15, 21, 27,…3,\ 9,\ 15,\ 21,\ 27,\ldots

Apart from 3 itself, every number there is divisible by 3\.

So a prime greater than 3 cannot occupy that lane either.

Six possibilities have become two:

1(mod6)1\pmod6

and

5(mod6).5\pmod6.

That is why the six-turn Prime Spring becomes so structured when the primes are emphasised.

There is a tempting mistake here.

The two surviving lanes are not prime lanes.

Consider 25\.

25≡1(mod6),25\equiv1\pmod6,

but

25=52.25=5^2.

Or 35:

35≡5(mod6),35\equiv5\pmod6,

but

35=5×7.35=5\times7.

The two lanes are merely compatible with larger primes. Divisibility by 2 and 3 has not ruled them out.

Other factors still can.

This is an important boundary:

the sieve can tell us where primes cannot occur; it does not tell us that everything surviving is prime.

\Open the modulo-6 residue view in the Prime Spring.\

And once that distinction is clear, the picture raises a much more interesting question.

The question no finite spring can answer

Look along the 5 modulo 6 lane.

We encounter primes such as

5, 11, 17, 23, 29, 41, 47, 53,…5,\ 11,\ 17,\ 23,\ 29,\ 41,\ 47,\ 53,\ldots

with composite numbers mixed between them.

Increase the range of the Prime Spring and we find more.

Increase it again and we find more still.

But however far we extend the experiment, the display remains finite.

That leaves open a possibility which no finite calculation can remove:

could there eventually be a last prime in this lane?

Perhaps the 5 modulo 6 lane continues producing primes for an unimaginably long time and then stops.

Perhaps one of the forty-eight prime-compatible lanes that we will later see modulo 210 eventually dries up.

No finite list can answer that.

A theorem can.

And that is the moment at which Dirichlet belongs in our story.

Dirichlet enters the picture

Peter Gustav Lejeune Dirichlet was born in Düren in 1805\. As a young man he studied in Paris, where he encountered leading figures of the French analytic tradition. He later became one of the most influential mathematicians working in Germany.

After Gauss died in 1855, Dirichlet took his chair at Göttingen. Riemann had already studied under Dirichlet in Berlin, and Dirichlet was an especially important influence on his mathematical style. When Dirichlet died in 1859, Riemann was appointed to the Göttingen chair. [1, 2]

The human chain is unusually neat:

Gauss → Dirichlet → Riemann.

There is a mathematical chain too.

Dirichlet had spent years thinking about primes inside the repeating kinds of sequence we have just discovered visually.

In 1837 he proved something extraordinary: under the right elementary condition, one of these arithmetic lanes can never run out of primes. [1]

Before stating his theorem, we need to translate our spring lane into the ordinary language in which the theorem is expressed.

ASCII PORTRAIT // Peter Gustav Lejeune Dirichlet

ASCII portrait of Peter Gustav Lejeune Dirichlet in three-quarter view, with a high forehead, swept-back hair, deep-set eyes, and a full beard, based on the supplied historical portrait source.

Peter Gustav Lejeune Dirichlet (1805–1859). True-text ASCII portrait based on the supplied reference portrait, reproduced in the 1889 first volume of G. Lejeune Dirichlet's Werke (BEIC collection).

From a lane to an arithmetic progression

Take our familiar lane

5, 11, 17, 23, 29,…5,\ 11,\ 17,\ 23,\ 29,\ldots

Another way to describe it is:

start at 5 and keep adding 6\.

So its terms are

5,5+6,5+2(6),5+3(6),…5,\quad5+6,\quad5+2(6),\quad5+3(6),\ldots

or

5+6n,n=0,1,2,…5+6n,\qquad n=0,1,2,\ldots

A sequence made by repeatedly adding the same amount is an arithmetic progression.

In general,

a, a+q, a+2q, a+3q,…a,\ a+q,\ a+2q,\ a+3q,\ldots

is an arithmetic progression beginning at aa, with common difference qq.

If we make a Prime Spring with qq integers per turn, those terms repeatedly occupy the same residue lane.

So two apparently different ideas have met:

  • a residue class tells us which repeating remainder family a number belongs to;
  • an arithmetic progression walks through that family in fixed steps.

But not every arithmetic progression has a chance of containing infinitely many primes.

Some are doomed before they begin.

The obstruction hidden inside a progression

Consider

6, 15, 24, 33, 42,…6,\ 15,\ 24,\ 33,\ 42,\ldots

The step size is 9\.

But every term is divisible by 3\.

Why?

Because both the starting value 6 and the step 9 share a factor of 3:

6+9n=3(2+3n).6+9n=3(2+3n).

The factor 3 is built into every term.

This suggests the right condition.

If the starting value aa and step qq share some factor greater than 1, the entire progression inherits that factor.

If they share no positive factor except 1, they are called coprime.

For example, 5 and 6 are coprime.

6 and 9 are not.

To express this compactly, mathematicians use the greatest common divisor.

The greatest common divisor of 18 and 30 is 6:

gcd⁡(18,30)=6.\gcd(18,30)=6.

Two numbers are coprime exactly when

gcd⁡(a,q)=1.\gcd(a,q)=1.

That small equation is the final piece we need.

Now Dirichlet's theorem can be stated without any unexplained machinery.

Dirichlet's theorem

Take the arithmetic progression

a, a+q, a+2q, a+3q,…a,\ a+q,\ a+2q,\ a+3q,\ldots

with q>0q>0.

If

gcd⁡(a,q)=1,\gcd(a,q)=1,

then the progression contains infinitely many prime numbers.

Equivalently, whenever the residue class a(modq)a\pmod q is coprime to qq, infinitely many primes satisfy

p≡a(modq).p\equiv a\pmod q.

This is Dirichlet's theorem on primes in arithmetic progressions. [3]

Our Prime Spring question now has an answer.

The 5 modulo 6 lane does not merely contain many primes in the finite part we happen to have drawn.

It can never run out.

There is no last prime congruent to 5 modulo 6\.

There is no last prime congruent to 1 modulo 6 either.

(The condition gcd⁡(a,q)=1\gcd(a,q)=1 is not a piece of technical decoration attached to the theorem. We have already seen why it has to be there. If aa and qq shared a divisor d>1d>1, every term a+nqa+nq would share that divisor. Dirichlet's theorem applies precisely when that elementary obstruction has disappeared.)

This is the first major boundary crossed by the article.

The spring gives us finite observations.

Dirichlet gives us an infinite conclusion.

And once we have the theorem, we can make the spring more complicated without losing our footing.

From two surviving lanes to forty-eight

Change the turn modulus from 6 to 30\.

Nothing happens to the underlying set of primes.

But the modular organisation changes because

30=2×3×5.30=2\times3\times5.

Any prime greater than 5 must therefore occupy a residue class coprime to 30\.

There are eight:

1, 7, 11, 13, 17, 19, 23, 29(mod30).1,\ 7,\ 11,\ 13,\ 17,\ 19,\ 23,\ 29 \pmod{30}.

How did we know there would be eight?

Euler's totient function counts the residue classes coprime to a modulus:

φ(30)=8.\varphi(30)=8.

Dirichlet then tells us something about each of those eight classes:

every one contains infinitely many primes.

\Open the modulo-30 residue view in the Prime Spring.\

Now change the modulus again:

210=2×3×5×7.210=2\times3\times5\times7.

This is especially revealing because any prime greater than 7 is automatically coprime to 210\.

Here,

φ(210)=48.\varphi(210)=48.

So 210 possible residue lanes collapse to just 48 that remain compatible with primes greater than 7\.

\Open the modulo-210 residue view in the Prime Spring.\

The other 162 are ruled out by divisibility by 2, 3, 5 or 7\.

Prime Spring's SIEVE view makes the elimination visible in stages:

210⟶105⟶70⟶56⟶48.210\longrightarrow105\longrightarrow70\longrightarrow56\longrightarrow48.

First eliminate multiples of 2\.

Then 3\.

Then 5\.

Then 7\.

What remains is an exact arithmetic skeleton underneath much of the striking appearance of the 210-turn spring.

\Step through the modulo-210 sieve view in the Prime Spring.\

(The number 48 describes residue classes, not primes. A class can survive every divisibility test imposed by 2, 3, 5 and 7 and still contain many composite numbers. “Prime-compatible” means only that these particular factors have not ruled it out.)

And Dirichlet now gives us not one infinite result, but forty-eight:

every one of those forty-eight coprime classes contains infinitely many primes.

That naturally leads to another question.

If they all continue forever, do they eventually receive an equal share?

Infinite is not the same as evenly shared

Look at the forty-eight surviving lanes over any finite range and their prime counts need not match.

One may be ahead.

Another may trail behind.

Those differences can remain substantial even when the range becomes very large.

Dirichlet's theorem does not say otherwise.

Its claim is about infinitude:

every coprime class keeps receiving primes.

A stronger asymptotic result — the Prime Number Theorem for arithmetic progressions — says that, for fixed qq, those classes eventually receive equal proportions in the long run.

If

π(x;q,a)\pi(x;q,a)

denotes the number of primes p≤xp\le x satisfying

p≡a(modq),p\equiv a\pmod q,

then for gcd⁡(a,q)=1\gcd(a,q)=1,

π(x;q,a)∼xφ(q)log⁡x.\pi(x;q,a) \sim \frac{x}{\varphi(q)\log x}.

An equivalent form uses the logarithmic integral Li⁡(x)\operatorname{Li}(x), which itself is asymptotic to x/log⁡xx/\log x:

π(x;q,a)∼Li⁡(x)φ(q).\pi(x;q,a) \sim \frac{\operatorname{Li}(x)}{\varphi(q)}.

So the φ(q)\varphi(q) coprime residue classes receive the same asymptotic share, even though their finite counts need not be equal.

This is not part of Dirichlet's 1837 theorem. The Prime Number Theorem for arithmetic progressions was first proved by Charles-Jean de la Vallée Poussin in 1896\. [3]

(The symbol ∼\sim is not an equals sign. It says that the ratio of the two quantities tends towards 1 as xx grows without bound. A finite Prime Spring can therefore show visibly unequal lane counts without conflicting in any way with the theorem.)

This distinction is worth pausing over.

We now have three different kinds of statement:

EXACT: modulo 210, exactly 48 residue classes are coprime to 210\.

FINITE: up to the current displayed range, a particular lane contains some exact number of primes.

ASYMPTOTIC: in the long run, all 48 coprime classes receive the same proportion.

Those are not interchangeable claims.

Keeping them separate is part of understanding what the mathematics actually says.

And it leaves us with a new problem.

We know what a particular lane looks like.

But how can analysis study that lane without drawing the spring at all?

How could an equation recognise one lane?

Suppose we work modulo 4\.

Every odd integer lies in one of only two residue classes:

1(mod4)1\pmod4

or

3(mod4).3\pmod4.

If we want to study those two lanes separately, we need some mathematical device capable of telling them apart.

Let us invent one before giving it a name.

Give every integer congruent to 1 modulo 4 the label

+1.+1.

Give every integer congruent to 3 modulo 4 the label

−1.-1.

And give every even integer the label

0.0.

So as the positive integers pass by,

1,2,3,4,5,6,7,8,…1,2,3,4,5,6,7,8,\ldots

their labels are

1,0,−1,0,1,0,−1,0,…1,0,-1,0,1,0,-1,0,\ldots

Call this labelling rule χ(n)\chi(n):

χ(n)={0,n even,1,n≡1(mod4),−1,n≡3(mod4).\chi(n)= \begin{cases} 0,&n\text{ even},\\ 1,&n\equiv1\pmod4,\\ -1,&n\equiv3\pmod4. \end{cases}

So far, this is just a repeating code for modular position.

Its real power appears when numbers are multiplied.

A filter that respects multiplication

Take 3 and 7\.

Both lie in the 3(mod4)3\pmod4 lane, so each receives label −1-1.

Their product is

3×7=21,3\times7=21,

and

21≡1(mod4).21\equiv1\pmod4.

The new label is therefore +1+1.

But

(−1)(−1)=+1.(-1)(-1)=+1.

The labels have multiplied in exactly the right way.

Try 5 and 7\.

The labels are

+1and−1.+1\quad\text{and}\quad-1.

Their product is 35, and

35≡3(mod4).35\equiv3\pmod4.

Again,

(+1)(−1)=−1.(+1)(-1)=-1.

What looks at first like a convenient colouring scheme has a much stronger property:

χ(mn)=χ(m)χ(n).\chi(mn)=\chi(m)\chi(n).

It respects multiplication.

Now we can give the object its name.

This is a Dirichlet character modulo 4.

In general, a Dirichlet character modulo qq is periodic modulo qq, completely multiplicative, and takes the value zero on integers that are not coprime to qq. [4]

(“Completely multiplicative” means that the rule χ(mn)=χ(m)χ(n)\chi(mn)=\chi(m)\chi(n) holds for every pair of positive integers m,nm,n. The compatibility with multiplication is the essential feature that will allow Euler's prime-product machinery to survive.)

We built the character because we wanted a lane detector.

Let us check that it really can act like one.

Turning the character into a lane selector

For an odd number nn,

χ(n)=+1\chi(n)=+1

means

n≡1(mod4),n\equiv1\pmod4,

while

χ(n)=−1\chi(n)=-1

means

n≡3(mod4).n\equiv3\pmod4.

So

1+χ(n)2\frac{1+\chi(n)}{2}

equals 1 on the 1(mod4)1\pmod4 lane and 0 on the 3(mod4)3\pmod4 lane.

Similarly,

1−χ(n)2\frac{1-\chi(n)}{2}

selects the 3(mod4)3\pmod4 lane.

The modular information has been turned into algebra.

For larger moduli, one character is generally not enough. But a whole collection of characters can be combined to isolate any residue class coprime to the modulus.

That ability comes from character orthogonality.

(For readers wanting the formal version, the characters modulo qq satisfy an orthogonality relation. For integers a,na,n coprime to qq, the indicator of one residue class can be written as 1 n≡a(modq)=1φ(q)∑χ mod qχ(a)‾ χ(n)\mathbf 1_{\,n\equiv a\pmod q}=\frac{1}{\varphi(q)}\sum_{\chi\bmod q}\overline{\chi(a)}\,\chi(n). This is the general form of the simple (1±χ(n))/2(1\pm\chi(n))/2 filters we just constructed modulo 4\. [4])

This is a significant step.

The Prime Spring separated the lanes geometrically.

Characters let mathematics separate them algebraically.

Now we need a way of feeding that filter into the analytic machinery that Euler had already built for primes.

Euler had learned how to hear the primes

Euler discovered that a certain infinite sum and a certain infinite product encode the same arithmetic.

For Re⁡(s)>1\operatorname{Re}(s)>1,

ζ(s)=∑n=1∞1ns=∏p(1−1ps)−1.\zeta(s) = \sum_{n=1}^{\infty}\frac1{n^s} = \prod_p \left(1-\frac1{p^s}\right)^{-1}.

The sum on the left runs through all positive integers.

The product on the right runs through the primes.

The bridge between them is unique prime factorisation: multiplying out the prime factors reconstructs every positive integer exactly once. [5]

This is the Euler product, one of the foundational ideas behind analytic number theory.

But Euler's original product treats primes collectively.

Our new problem is more selective.

We want to retain information about which modular lane a prime belongs to.

The character gives us precisely such a label.

So the next move is natural:

put that label into the series.

From a lane filter to an L-function

Instead of weighting every integer simply by

1ns,\frac1{n^s},

weight it by

χ(n)ns.\frac{\chi(n)}{n^s}.

Then sum:

L(s,χ)=∑n=1∞χ(n)ns.L(s,\chi) = \sum_{n=1}^{\infty} \frac{\chi(n)}{n^s}.

This is a Dirichlet L-function.

In modern language, Dirichlet L-functions are the analytic objects built from Dirichlet characters. They arose from Dirichlet's work on primes in arithmetic progressions. [6]

Notice the route we took to get here.

We did not begin by announcing an L-function.

We began with a visible arithmetic problem:

how can we distinguish one prime-compatible lane from another?

We created a repeating modular label.

We discovered that the label respected multiplication.

Only then did we place it inside an infinite series.

And because multiplication survived, something remarkable happens next.

Euler's product survives too.

The prime product remembers the lane

For Re⁡(s)>1\operatorname{Re}(s)>1,

L(s,χ)=∏p(1−χ(p)ps)−1.L(s,\chi) = \prod_p \left( 1-\frac{\chi(p)}{p^s} \right)^{-1}.

Compare this with the zeta Euler product:

ζ(s)=∏p(1−1ps)−1.\zeta(s) = \prod_p \left( 1-\frac1{p^s} \right)^{-1}.

The architecture is almost identical.

But in the L-function, every prime carries an extra piece of arithmetic information:

χ(p).\chi(p).

For our modulo-4 example, that value tells us whether an odd prime lies in the 1(mod4)1\pmod4 lane or the 3(mod4)3\pmod4 lane.

The prime product has learned to remember residue classes. [6]

That is the central conceptual achievement.

Euler's analytic machinery for primes has been equipped with an arithmetic filter.

This is why the character had to respect multiplication. Without that property, the clean factorisation over primes would fall apart.

And now we have reached the genuinely deep part of Dirichlet's argument.

Why s=1s=1 matters

It would be easy to make the proof sound deceptively simple at this point.

We have characters.

We have L-functions.

We have Euler products.

Surely we can now simply read off that every coprime residue class contains infinitely many primes.

But this is exactly where the serious analysis lives.

One crucial fact is that for every non-principal Dirichlet character,

L(1,χ)≠0.L(1,\chi)\ne0.

DLMF records this explicitly. [6]

Why should a statement about the value of an analytic function at s=1s=1 tell us something about infinitely many primes in one arithmetic progression?

The broad architecture is this.

Characters allow us to combine prime sums in such a way that a chosen residue class is isolated.

The principal character supplies a divergent contribution near s=1s=1, reflecting the same kind of logarithmic prime accumulation that appears behind Euler's proof of the infinitude of primes.

The other characters must not introduce a competing zero at s=1s=1.

Their non-vanishing keeps those additional contributions controlled.

The isolated prime sum for the chosen residue class is then forced to grow rather than settle to a finite value.

A finite collection of primes could not do that.

Therefore the residue class must contain infinitely many primes.

(This is deliberately an architectural explanation rather than a disguised proof. A complete proof requires careful treatment of characters, logarithms of Euler products, convergence and the non-vanishing theorem at s=1s=1. The important conceptual point here is that the elementary residue-class question has been converted into a statement about analytic functions.)

We have now travelled a surprisingly long way from six lanes on a spring.

But the route has been continuous.

Remainders led to residue classes.

Residue classes led to arithmetic progressions.

Arithmetic progressions led to Dirichlet's theorem.

Trying to isolate one progression led to characters.

Multiplicative characters led to L-functions.

And L-functions lead directly towards Riemann.

Dirichlet and Riemann: a human and mathematical bridge

Euler had connected infinite series to prime products.

Dirichlet modified that machinery so that it could distinguish arithmetic progressions.

Riemann then transformed the study of the zeta function by treating it as a function of a complex variable and making its zeros central to the study of prime distribution. [7]

The historical overlap is unusually close.

Riemann studied under Dirichlet in Berlin, and Dirichlet strongly influenced him. Dirichlet took Gauss's Göttingen chair in 1855; after Dirichlet's death in 1859, Riemann was appointed to it. [2]

The mathematical relationship is just as important.

The zeta function is not an isolated curiosity.

It belongs to a broader world of functions whose coefficients encode arithmetic information and whose prime products make that information accessible to analysis.

Dirichlet L-functions are among the earliest and most fundamental examples.

Once we see zeta as one member of a larger analytic family, an obvious question arises.

If Riemann's zeros matter for the global distribution of primes, what do the zeros of these lane-sensitive L-functions tell us?

From RH to GRH

The Riemann Hypothesis concerns the non-trivial zeros of

ζ(s).\zeta(s).

It predicts that all of them lie on the critical line

Re⁡(s)=12.\operatorname{Re}(s)=\frac12.

For primitive Dirichlet characters, the corresponding L-functions have non-trivial zeros in a critical strip. Their distribution is intimately connected with the distribution of primes in arithmetic progressions. [6]

The corresponding larger conjecture is the Generalized Riemann Hypothesis.

For primitive Dirichlet characters, GRH predicts that the non-trivial zeros of their L-functions lie on

Re⁡(s)=12.\operatorname{Re}(s)=\frac12.

(Every Dirichlet character is associated with a primitive character of a conductor dividing the modulus. The corresponding L-function differs only by finitely many Euler factors, so zero questions are conventionally reduced to the primitive case. [6])

That makes the progression from RH to GRH much less mysterious than the name can initially suggest.

The zeta function has forgotten which arithmetic lane a prime occupies.

Dirichlet L-functions remember.

RH asks for extremely precise structure in the zeros of zeta.

GRH asks for the analogous structure across the larger family of arithmetic-sensitive L-functions.

(GRH is not needed for Dirichlet's theorem. Dirichlet had already proved that suitable arithmetic progressions contain infinitely many primes decades before Riemann formulated his hypothesis. GRH concerns much finer control over the distribution and fluctuations of primes within those progressions.)

And with that, we can return to where we began.

Not to an equation, but to the spring.

What the Prime Spring can — and cannot — tell us

At modulus 210, Prime Spring can tell us exactly that

φ(210)=48.\varphi(210)=48.

That is an exact arithmetic statement.

It can count exactly how many displayed primes occupy each of those forty-eight lanes up to the selected range.

That is finite data.

It may show one lane ahead of another.

That is a finite-range observation.

Dirichlet's theorem tells us that every one of those forty-eight lanes contains infinitely many primes.

That is an infinite theorem.

The Prime Number Theorem for arithmetic progressions tells us that the lanes receive equal asymptotic proportions.

That is a long-run theorem.

GRH, if true, would give much finer information about errors and fluctuations around such expected distributions.

That is a conjectural statement.

The beauty of the Prime Spring is not that it somehow proves all of these things visually.

It is that it places several layers of number theory in the same field of view while allowing us to distinguish them.

The visible lanes come from modular arithmetic.

Their elimination comes from divisibility.

Their surviving count comes from Euler's totient function.

Their inexhaustible supply of primes comes from Dirichlet.

Their long-run balance belongs to deeper analytic prime-distribution theory.

Characters let us isolate them algebraically.

L-functions let analysis study them.

Their zeros lead towards GRH.

These are different statements answering different questions.

The structure becomes more interesting when we keep them separate.

A finite window onto an infinite statement

Imagine that one residue lane contains 300 primes in the visible range.

Increase the range and perhaps it contains 700\.

Increase it again and it contains 2,000.

Then 10,000.

At every stage, the computation has told us something exact about a finite collection of integers.

But however large the number becomes, it has not answered this:

must another prime eventually appear?

A finite experiment can never cross that logical boundary.

There is always another point beyond the last point examined.

That difference between computation and proof is easy to state but profound in practice.

A computation can travel extraordinarily far.

A theorem can speak about every distance beyond it.

This is why the Prime Spring and Dirichlet's theorem belong together so naturally.

The spring makes the question visible.

The theorem answers the part that sight cannot reach.

Follow the thread

The main stepping stones behind the article now have their own glossary entries.

Modular arithmetic develops modulo, modulus and congruence from remainder calculations.

Residue class explains the repeating families that become lanes in Prime Spring.

Arithmetic progression gives the ordinary sequence language behind a lane such as

5, 11, 17, 23,…5,\ 11,\ 17,\ 23,\ldots

Coprime explains what it means for two integers to have no shared positive divisor greater than 1\.

Greatest common divisor develops the condition

gcd⁡(a,q)=1.\gcd(a,q)=1.

Euler's totient function counts the coprime residue classes — including the forty-eight prime-compatible classes modulo 210\.

The Prime Spring lets you explore the geometry directly.

For the wider story of prime density and counting, see How are the prime numbers distributed?.

For the human and mathematical background to the analytic machinery, continue with Who was Leonhard Euler? and Who was Bernhard Riemann?.

The one sentence to remember

The Prime Spring can show us prime after prime arriving in an arithmetic lane.

We can extend the range.

Then extend it again.

But every picture we draw remains finite.

So the picture can say:

“Here are more primes.”

Dirichlet says something no finite picture ever can:

“There will never be a last one.”

References

  1. J J O'Connor; E F Robertson. “Lejeune Dirichlet (1805–1859).” MacTutor History of Mathematics (2000). Source
  2. J J O'Connor; E F Robertson. “Bernhard Riemann (1826–1866).” MacTutor History of Mathematics (1998). Source
  3. NIST Digital Library of Mathematical Functions. “§27.11 Asymptotic Formulas: Partial Sums.” NIST Digital Library of Mathematical Functions, Release 1.2.8 (2026). Source
  4. NIST Digital Library of Mathematical Functions. “§27.8 Dirichlet Characters.” NIST Digital Library of Mathematical Functions, Release 1.2.8 (2026). Source
  5. Leonhard Euler. “Variae observationes circa series infinitas (Various observations about infinite series).” Commentarii academiae scientiarum Petropolitanae / Euler Archive 9 (1744). 160–188. Source
  6. NIST Digital Library of Mathematical Functions. “§25.15 Dirichlet L-functions.” NIST Digital Library of Mathematical Functions, Release 1.2.8 (2026). Source
  7. Bernhard Riemann. “Ueber die Anzahl der Primzahlen unter einer gegebenen Grösse (On the Number of Primes Less Than a Given Quantity).” Monatsberichte der Berliner Akademie / Clay Mathematics Institute manuscript collection (1859). Source

Glossary connections

  1. [..]Arithmetic progressionGLOSSARY · STANDARD MATHEMATICS
  2. [..]CoprimeGLOSSARY · STANDARD MATHEMATICS
  3. [..]Euler productGLOSSARY · RIEMANN HYPOTHESIS
  4. [..]Euler's totient functionGLOSSARY · STANDARD MATHEMATICS
  5. [..]Greatest common divisorGLOSSARY · STANDARD MATHEMATICS
  6. [..]Modular arithmeticGLOSSARY · STANDARD MATHEMATICS
  7. [..]Residue classGLOSSARY · STANDARD MATHEMATICS