Actually Rain terms of your OP and knowing when to stop mixing and start mastering.  The answer to that is just keep your mixing hat on and go for a very well balanced mix that just sounds great.
 
Keep the monitoring levels in your room up so you are not having any problems hearing your great mix. (make it sound like a mastered level. You really need an SPL meter in the mix stage, I could not live without mine)  By turning your mix up you are removing the compression and limiting part of the mastering almost.  Leaving the EQ.  And if you get your mix sounding stellar there won't be much EQ work to do in mastering either.  Just some tidying up of the very ends of the spectrum and some fine tuning of the mids usually.
 
You cannot make mastering decisions after bashing your ears for 8 hours mixing either. Put the mix away for a week and listen to it quietly here and there and note any possible changes. To my ears importing a premastered mix into a totally fresh (week later) mastering session and applying EQ, compression and limiting always sounds better there than trying to do it all at the time of the mix.  Not to mention you won't need all that extra heavy CPU work required if you don't slam a bunch of processors over a mix.
 
Sometimes a little gentle two buss compression conditioning can be nice but I don't consider that mastering. I usually ensure if I do that, the rms (VU) levels going in and out of that process are the same.
 
Learn to live and love the premastered sound. A lot of people think they only have to go 80% of the way with the mix and the mastering will do the rest. I dont agree with that approach. I think it is far better to go 98% of the way with the mix and then you will only have some nice light and simple mastering processes to do which WONT destroy your mix at all but make it just that little bit nicer.  That is the key.
 

whats up ?…...did I miss something ? ……..where's the song / mix ? 
 
how can the idea of mastering a song or a mix apply to creating a working mix for a vocalist to sing over ?
 
Yes ? No ?
 
I can see being concerned about keeping the gain staging right , carving frequencies , making space in a mix , the proper use of effects , and leaving a little headroom to work with ... 
 
in my ever so humble opinion  ,
the song is only a work in progress until the vocals are done and the song is completely mixed and arranged to perfection ,  
 
THEN YOU MASTER   
 
sheessshhh 
 
Kenny

Kenny - the song isn't uploaded anywhere, except on our local network here. ;)
 
Though it makes no sense to even just mix a song before the vocals are recorded, I'm using it as an opportunity to learn.
 
Mastering... That's something I'm not trying in this life time. :P I barely have enough time left to learn what I want to learn in this life, and maybe finally be good at it. There's no way I'm wasting time on such a hopeless pursuit as mastering.
 
But this one mix, by its own instrumental self, seemed to be clearly closer to the mark than most of what I've ever done, and with so little as 3 minor bump or cuts with an EQ was even closer.
 
So I did get back to the mix itself and tried to focus on carving out a bit of the same frequencies. Even though my EQ'ing on the stereo buss had been pretty subtle, I opted to go for something even more subtle with the individual tracks.
 
Turning the EQ on the main out buss back on after I'd made the adjustments to individual tracks now made everything sound exaggerated, telling me that I'd probably did an adequate job doing the modifications to the actual tracks.
 
So if anything, the whole process just allowed me to figure out more quickly which ranges I should focus on.
 
As for the rest, however it goes... Whatever we release when we release it I don't want to sound too conventional and commercial. As far as I'm concerned, it should be a lot more raw and dry and not overproduced. No auto-tune - not needed anyway - no brickwall limiting, no TR-808 bass drum sample.
 
Her first album was sterile sounding - they even managed to make real strings sounded fake. She deserves better. My wife is a 100% genuine rocker - and I don't mean a cutie wearing old Judas Priest shirts brought to her by a stylist... I mean the ass-kicking real thing. The production should reflect that attitude.
 

Jeff Evans
Actually Rain terms of your OP and knowing when to stop mixing and start mastering.  The answer to that is just keep your mixing hat on and go for a very well balanced mix that just sounds great.
 
Keep the monitoring levels in your room up so you are not having any problems hearing your great mix. (make it sound like a mastered level. You really need an SPL meter in the mix stage, I could not live without mine)  By turning your mix up you are removing the compression and limiting part of the mastering almost.  Leaving the EQ.  And if you get your mix sounding stellar there won't be much EQ work to do in mastering either.  Just some tidying up of the very ends of the spectrum and some fine tuning of the mids usually.
 
You cannot make mastering decisions after bashing your ears for 8 hours mixing either. Put the mix away for a week and listen to it quietly here and there and note any possible changes. To my ears importing a premastered mix into a totally fresh (week later) mastering session and applying EQ, compression and limiting always sounds better there than trying to do it all at the time of the mix.  Not to mention you won't need all that extra heavy CPU work required if you don't slam a bunch of processors over a mix.
 
Sometimes a little gentle two buss compression conditioning can be nice but I don't consider that mastering. I usually ensure if I do that, the rms (VU) levels going in and out of that process are the same.
 
Learn to live and love the premastered sound. A lot of people think they only have to go 80% of the way with the mix and the mastering will do the rest. I dont agree with that approach. I think it is far better to go 98% of the way with the mix and then you will only have some nice light and simple mastering processes to do which WONT destroy your mix at all but make it just that little bit nicer.  That is the key.
 

Words of wisdom, and what I was hoping would still be true.
 
You know, one of the reasons I praise Logic is its ability to selectively import data from another project on a very granular basis. I even did a short vid about that feature I posted here a couple of years ago. That feature makes it very easy to start afresh without having to rebuild everything from scratch. Markers, tracks, busses, routing, busses, effects - you pick and chose what you want to import.
 
Also makes it a breeze to create an alternative mix.
 
I do use some very gentle compression on the master buss - 2:1 or 3:1, usually just shaving off 2 or 3 db's and glueing things up together.
 
Ideally, I don't want what we do to sound like what's out there. Personally, I'd seriously consider bypassing mastering altogether. I don't like the sound of mainstream music nowadays.
 
Furthermore, iTunes and more and more streaming services make automatic level adjustments, so you actually will deliver more punch and apparent loudness if you down't squeeze every last db out of your limiter. I see that as an opportunity to start making music that sounds good again.
 
For that, getting the mix 98% there is still essential though.
 

Rain
Kenny - the song isn't uploaded anywhere, except on our local network here. ;)
 
Though it makes no sense to even just mix a song before the vocals are recorded, I'm using it as an opportunity to learn.
 
Mastering... That's something I'm not trying in this life time. :P I barely have enough time left to learn what I want to learn in this life, and maybe finally be good at it. There's no way I'm wasting time on such a hopeless pursuit as mastering.
 
But this one mix, by its own instrumental self, seemed to be clearly closer to the mark than most of what I've ever done, and with so little as 3 minor bump or cuts with an EQ was even closer.
 
So I did get back to the mix itself and tried to focus on carving out a bit of the same frequencies. Even though my EQ'ing on the stereo buss had been pretty subtle, I opted to go for something even more subtle with the individual tracks.
 
Turning the EQ on the main out buss back on after I'd made the adjustments to individual tracks now made everything sound exaggerated, telling me that I'd probably did an adequate job doing the modifications to the actual tracks.
 
So if anything, the whole process just allowed me to figure out more quickly which ranges I should focus on.
 
As for the rest, however it goes... Whatever we release when we release it I don't want to sound too conventional and commercial. As far as I'm concerned, it should be a lot more raw and dry and not overproduced. No auto-tune - not needed anyway - no brickwall limiting, no TR-808 bass drum sample.
 
Her first album was sterile sounding - they even managed to make real strings sounded fake. She deserves better. My wife is a 100% genuine rocker - and I don't mean a cutie wearing old Judas Priest shirts brought to her by a stylist... I mean the ass-kicking real thing. The production should reflect that attitude.
 

 
Hello Rain , 
 
I got you now ….I missunderstood … error on my part 
 
good luck , may your musical dreams come true ,
 
Kenny 

Ok this has been done to death and I have come late to my own funeral.
 
There are lot's of caveats to what the guys have said above, and indeed to your question.  Mastering as it used to be, does not really exist anymore.  The process of mastering was invented in the late 50's if I remember correctly (I'd have to look it up) to balance vinyl...notice the word vinyl.  Up until then (working from memory) the finished recording was mostly just pressed to the vinyl with no real tonal adjustments.  As recordings became more complex i.e The Beatles a whole new process had to be invented to make sure those sounds could be re-produced on a mono yea mono hi-fi system.  And again this progressed throughout the 70's.  Till tape emerged as the pre-dominate format only to be quickly replaced by CD, then funnily enough not long after Mp3 emerged the winner.
 
Now what am I on about, mastering to each of these different formats takes a different skill set.  I am not a vinyl mastering engineer...don't want to be.  I'm not sure how many of the above people can actually master to vinyl...maybe we should ask "can any of you master to vinyl?"  Now mastering to vinyl is a handy skill set because it's a boutique industry with 1 million approx. sales in the UK last year (yea it's actually not that many) if you have the skills and equipment it might be a nice stocking filler.
 
For the rest of us however, we are firmly rooted in the future, I will never release to vinyl and probably never to CD.  However if you were to release to CD again you might need a mastering engineer to help if you don't know the procedure... i.e. redbook audio cd with no more than (I can't remember off top of my head) insert figure burning gremlins and no music file going over 0.2.  You get the drift, then there is registering the tracks with gracepoint (again memory) oh and a basic understanding of dynamics...the loudness wars and how different digital convertors behave.  You can see why a digital mastering engineer might be handy...so that means Danny and Jeff are pretty safe.
 
However, if you are going to release the bog standard Mp3, then you don't really have to worry about all that bollocks.  And truth be told, we are in a totally different era to Danny and Jeff, and this is not a slight at them.  But in all honesty their recording or sonic arts paradigm (now I am being a pretensions wanka) is now consigned to history.  Once upon a time the Jeff's and Danny's were the goto boffins, and they deserved the title of boffin just as some of the best engineers enjoyed that title think Emerick, Norman Smith, Glyn John (The Beatles again) (whose son worked on McCartney's record) and of course Alan Parsons...lol got to love Alan Parson's, the ego on this guy was so huge boffin was not enough he wanted to be a rock star to and he made it...sort of.
 
Now to answer your question, sorry I have a terrible propensity to take the long road...the walk is more enjoyable.  I mix and master as I go, and I get results within range.  Yes fellas I have put up some ****e, because I don't believe in hiding behind skirts but after almost 15 years of perfecting a formula all my latest mixes are within range.
 
This is a very different paradigm to the one that Jeff and Danny are talking about, today we are no longer engineer's or boffins although they still exist...Danny and Jeff and they will continue to exist although less and less for music and more for broadcast.  Today as long as you follow a formula, you can mix and master all at the same time, so that when you complete a track within a hour of completion your master is ready.  Even with tired ears.  As long as you have followed the formula and mixed as you go, then what I am saying is true.  Gosh I'm sounding like an infomercial.
 
I can say this with some authority, as I have mentioned I have spent 15 years in the wilderness perfecting the formula...yes I am bat **** crazy.  I have completed almost three degrees in our field (third one about to finish next month with a top mark of 84 so far, on the history of sonic arts) and "with a little luck" (ah Paul ****ing McCartney still going strong...although he is going senile Jayz the best concert he ever attended) a PHD offer next year.  Although with this caveat, I make the above claim, I am based in the land of theory, although all my music is up on soundcloud so you can judge yourself, I have had music I have written and produced used on three different TV shows here in Oz too.  But I have never had a hit record or released a record.  So take my advice as you will.
 
To conclude, we are in a different era of music/sonic production.  The mastering engineer does exist but is a bit of an anachronism, they're not necessary unless your going to release to vinyl...and who is going to release to vinyl and I point to all the pretentious wankas out there.  You can mix and master all at the same time, and you should mix and master at the same time, the line does not exist anymore.  Today's sonic composer "is all things as he is no things" (paraphrasing Malcom McDowell...Caligula 1980 ;)) as long as you have the skill set, there is no reason why you can't paint beautiful mixes and masters.  Sonic Digital Impressionism or something along those lines.  This is what the 60's masters were, although they needed a team to create their sonic masterpieces...today we can do it with as one man or woman with a machine that fit's into the palm of our hands.  Ah it's a funny time.  But if anyone wants the formula, or the treatise or even the manifesto on Sonic Digital Impressionism, one day I will sit down and write it out for you.
 
Finally this post was not to knock Danny or Jeff, but to offer a third way.  Their way was correct and is still correct but it's fast becoming history...everything becomes history even The Beatles...even you and me.  So Danny and Jeff I'm just offering my answer to the above question, which is the line no longer exists between mixing and mastering.  We are now sonic painters, capturing the light within the mundane ;) Van Gough would be proud.     
 
Ben  

Why high pass 40 Hz if you may not need to?
Why mess with 240 or 5k "just because"?
 
Oh Danny is right here, but what he is saying use your ear and I would also add use your spectrum analyser, it can tell you when your ears can't.  Within the formula there is room for movement.
 
Ben

As I said earlier: 
"The low cut at 40 seems arbitrary, and to a certain degree it is. I guess I was partially influenced to take the decision by some (maybe questionable) common wisdom. Other key factors were the genre (rockabilly/swing) and overall production feel I had in mind, and personal taste.
 
Furthermore, my set up doesn't allow me to make an informed judgement on what's going on down there. The way the little mono set up in the gym reacted also gave me the impression that there was too much going on in the low frequencies.
 
240 dip: most of the stuff I listen to seem to have some kind of dip in that region compared to my own mixes. So it's not just the reference mix, which, to be honest, I didn't really attempt to duplicate. As a matter of fact, I simply played back a few passages to give me a feel, but didn't really analyze it."
 
So that's basically following what my ears are telling me, and a certain gut feeling, as well as my very own preferences.
 
As for the Analyzer, I'm fortunate enough to have access to one right into my default EQ - and a LARGE one at that (shown at 111% below but can be enlarged as needed). That's the only EQ I need and use. So anytime I EQ anything, I can always have a quick look at the spectrum, pre and post EQ.
 

 
 

 
That's how I work these days, the screen never lies in that regard.  I can see from that screen shot that there is a nasty bump at around 70ish Hz, and one that could be tackled in a couple of ways, either use a low-cut filter and do a general clean or maybe just a shelving filter and gently massage it out, actually there are four ways, or a general EQ and carve it out or finally a multi-band compressor.  But if you are in the mixing stage and about to go into the mastering stage...the question is what is causing this issue?  The mix from that screen shot is not very even.
 
Ben 

I think you hire a Mastering Engineer depending on the quality of the end product you are expecting to obtain.  If you are a professional musician wanting to put out polished professional work, you hire people like Danny and Jeff that have REAL skins on the wall.  You would probably do best to track in a true studio environment as well.....with real instruments.  If your a home studio artist posting 256 mp3 files, you can do fine small mastering your own. 
 
I feel very confident that if either Danny or Jeff were to Master a project that I did, it would turn out far better than what I could do myself, even though I have been doing this for 10 years as well.  It's called experience, quality equipment and tested environment.  There is a reason that clientele still go to the proven, for the professional end product.