Why atoms change size across the table
The atomic radius follows two main directions. Across a period from left to right, atoms get smaller. Down a group, they get larger. The reason is straightforward but easy to mess up when you first learn it. Effective nuclear charge is the real driver here. As you move across a period, protons are added to the nucleus while electrons fill the same principal energy level. The shielding from inner electrons stays roughly constant because no new inner shells are introduced. Each extra proton pulls the electron cloud tighter. So sodium is notably bigger than chlorine even though they sit in the same row. Down a group, the principal quantum number increases. Each step down adds a new electron shell. The outer electrons sit further from the nucleus. Shielding from inner shells more than compensates for the growing nuclear charge. That is why cesium is enormous compared to lithium.
The tricky part is that atomic radius is not a single fixed number for any element. It depends on whether you are measuring covalent radius, metallic radius, or van der Waals radius. Covalent radius comes from half the bond length in a diatomic molecule or a covalent network. Metallic radius comes from the distance between adjacent atoms in a metal lattice. Van der Waals radius measures the distance between non-bonded atoms in adjacent molecules. These values diverge significantly for the same element, especially for larger atoms where the electron cloud is diffuse. I spent too long early on mixing these up and then wondering why my calculated densities didn't match literature values. Once I started pulling radius data from the specific context matching my problem rather than just grabbing a single number from a table, everything else fell into place. Here is a common pitfall that people walk into repeatedly. Transition metals across a period do not shrink as dramatically as main-group elements. From scandium to copper, the atomic radius decreases only slightly, and then it barely changes at all through zinc. The reason is that the added electrons go into the d subshell rather than the outermost s shell. The d electrons provide some shielding, so the effective pull on the outer electrons is dampened. You cannot simply project the steep main-group trend onto the d-block and expect accurate results.
Another thing that catches people out is the lanthanide contraction. After lanthanum, you fill the 4f subshell across the lanthanides. The 4f electrons shield poorly, so the increasing nuclear charge pulls the outer electrons inward more than expected. By the time you reach hafnium through mercury, the atomic radii are nearly identical to their 5d counterparts above them in the same groups. Zirconium and hafnium, for example, have almost the same atomic radius despite being in different periods. This is why separating them chemically is so difficult and why their physical properties overlap so much. I ran into this directly when I was modeling crystal structures for a project involving tantalum and hafnium carbides. I assumed the lattice parameters would scale cleanly with atomic radius the way they do for simpler compounds. They did not. The lanthanide contraction had compressed hafnium to nearly the same size as zirconium, which threw off my initial estimates. I ended up pulling experimental lattice data from the Joint Committee on Powder Diffraction Standards files instead of relying on trend-based predictions. It saved me from weeks of recalibrating the model. The trend breaks down in a few other predictable ways. Noble gases are often listed with van der Waals radii rather than covalent radii because they do not form stable covalent bonds under standard conditions. Comparing a noble gas van der Waals radius to a halogen covalent radius will make the noble gas look absurdly large. You have to be explicit about which radius type you are using whenever you discuss this.
Get the Full Details
Relativistic effects also become noticeable in the heaviest elements. Gold appears yellow partly because relativistic contraction of the 6s orbital shifts its absorption spectrum. Mercury is liquid at room temperature because relativistic stabilization of the 6s electrons weakens metallic bonding. These effects cause deviations from the expected trend in the sixth period that pure electrostatic reasoning does not capture. If you need specific numbers, the most reliable source I have found is the CRC Handbook of Chemistry and Physics. It lists multiple radius values for each element with clear citations to the original measurements. Some online tables present a single value per element, which looks clean but hides the ambiguity between covalent, metallic, and van der Waals radii. When precision matters, go to the handbook or the original crystallography papers. The periodic trend itself is useful as a first approximation. For quick comparisons within the main group, it works very well. Predicting whether sodium chloride or magnesium oxide has a higher lattice energy based on ionic radii comes out right every time. But for transition metals, lanthanides, or anything involving heavy elements, the simple trend is not enough. You need actual measured data for the specific radius type relevant to your application.
Data tables are available from the NIST Chemistry WebBook and the CRC Handbook. Neither requires a subscription for basic radius values. I usually download the full element-by-element tables into a spreadsheet and flag which radius type applies to each entry so I never mix them up again.
