I think there are multiple misunderstandings here - like what a spectrum actually is, what it represents, that you can have absorption and emission spectra at the same time, the contributing factors to the intensity vs wavelength curves, how spectral lines form, stuff like that.
It would be useful to know what level this question is being asked at.
ehild basically shows how you get energy bands in solids: they are extremely close energy levels - transitions between very close levels will look continuous unless you try quite hard to look carefully ... and you get HUP limits too, so the "discrete" energy levels are actually a range of possible values. Get the energy levels close enough together and they overlap.
You can get transitions between two bands though ... the range of available energies is what gives an LED spectra it's broad shape (I mean without the lens or any coatings that manufacturers add later).
There's another effect - where you have transitions from the continuum: i.e. unbound states.
(Simplifying somewhat...)
When an electron is no longer bound to an atom, it may still be contained within the apparatus.
We can model the apparatus as an infinite square well (say) so there are still discrete energy levels to transition from... it's just that they are much closer together than the atomic energy levels. i.e. The atom has dimensions of order 1A, while the apparatus likely has dimensions of order of 1cm = 100000000A so we can expect the energy levels to be 100-million times closer together. The prev-mentioned HUP limit works here too.
In a plasma - for instance - the media is so hot that the atoms are totally ionized. The only available transitions will be between these continuum states rather than atomic states ... so the only effect on the spectra will be the distribution of particles between these states ... which is pretty much the definition of "temperature".
Therefore we'd expect the spectra of a plasma to be continuous and to depend only on it's temperature.
It's a pretty big subject.
Which is why it is useful to narrow it down.
I suspect OP is just asking how come you get a full rainbow when you put light from a carbon arc through a prism, but not when you put light from hydrogen or mercury ... the answer will be that the prism presumably used was just not an accurate enough instrument to separate out the spikes in the carbon-arc spectra ... presumably it wouldn't be able to separate the orange peaks in the sodium spectra either.