The first tag to appear within each <app> apparatus entry must always be <lem>, which will include the exact text we publish in our edition:

<l n="54">
	<app>
		<lem>Eumenidas</lem>
		<rdg wit="#B #δ #L #μ #ν #O #Q">Eumenides</rdg>
	</app>
	 perhibetur aquis;
	 <app>
		<lem>huc</lem>
		<rdg wit="#D #δ #μpc #N #νac #O #Qpc #t">hic</rdg>
		<rdg wit="#Qac">huic</rdg>
	</app>
	 mergere
	 <app>
		<lem>suetae</lem>
		<rdg wit="#P">sueta</rdg>
		<rdg wit="#Baehrens #Hall">suetas</rdg>
	</app>
</l>

In the example you can see that there are three apparatus entries in the same verse. The content of the three <lem> will be that which appears in the body of the published text, which will read, in this case, "Eumenides perhibetur aquis; hoc mergere suetae".

We do not add the attribute @wit to the <lem> tag to indicate which manuscript or publisher corresponds to this reading, since we always take the reading from our reference edition. If we need to indicate these witnesses for the lemma, a new <rdg> must be created to contain this information (cf. <rdg>).